Detoxified O-Specific Polysaccharide (O-SP)–Protein Conjugates: Emerging Approach in the Shigella Vaccine Development Scene

Shigella is the second most common cause of moderate to severe diarrhea among children worldwide and of diarrheal disease-associated mortality in young children in low-and middle-income countries. In spite of many years of attempts to develop Shigella vaccines, no licensed vaccines are yet available. Injectable conjugate vaccines made of the detoxified lipopolysaccharide (LPS) of S. flexneri 2a, S. sonnei, and S. dysenteriae type 1 covalently bound to protein carriers were developed in the early 1990s by John B. Robbins and Rachel Schneerson at the US National Institutes of Health. This approach was novel for a disease of the gut mucosa, at a time when live, rationally attenuated oral vaccine strains that intended to mimic Shigella infection and induce a protective local immune response were extensively investigated. Of keystone support to Shigella glycoconjugates development were the findings of a strong association between pre-existent serum IgG antibodies to S. sonnei or S. flexneri 2a LPS and a lower risk of infection with the homologous Shigella serotypes among Israeli soldiers serving in field units. In view of these findings and of the successful development of the pioneering Haemophilus influenzae type b conjugate vaccines, it was hypothesized that protective immunity may be conferred by serum IgG antibodies to the O-Specific Polysaccharide (O-SP) following parenteral delivery of the conjugates. S. sonnei and S. flexneri 2a glycoconjugates induced high levels of serum IgG against the homologous LPS in phase I and II studies in healthy volunteers. The protective efficacy of a S. sonnei detoxified LPS-conjugate was further demonstrated in field trials in young adults (74%) and in children older than three years of age (71%), but not in younger ones. The evaluation of the Shigella conjugates confirmed that IgG antibodies to Shigella LPS are correlates of protection and provided solid basis for the development of a new generation of glycoconjugates and other injectable LPS-based vaccines that are currently in advanced stages of clinical evaluation.


Introduction
Shigella is a leading cause of moderate to severe diarrhea and continues to be associated with a considerable burden of disease across the globe. Shigellosis is highly endemic in lowand middle-income countries (LMICs) where approximately 250 million new cases occur yearly with over 212,000 fatal cases of whom~63,000 among children less than five years old [1]. Children with shigellosis have an increased risk of persistent diarrhea, malnutrition, and linear growth failure [1][2][3]. Shigella is also responsible for around 2 million cases of shigellosis in high-income countries [1,4] where at-risk populations include children below five years of age, living in conditions of crowding [5,6], soldiers serving under field conditions, men who have sex with men, and travelers to LMICs [1,7,8].
An emerging global increase in the antimicrobial resistance of Shigella narrows the antibiotic treatment options, especially in young children [9][10][11].

Shigella Vaccine Development and the Emergence of Injectable Shigella Conjugate Vaccines into the Scene
Shigella vaccine development strategies of the last 60 years included two major approaches to deliver the protective Shigella LPS antigen to the host immunocompetent cells: live-attenuated vaccine strains and inactivated Shigella vaccine candidates (whole cell or subunit).

Live-Attenuated Shigella Vaccines
Orally delivered, live-attenuated Shigella vaccines have been developed with the intention to mimic natural Shigella infection and induce a protective local immune response against shigellosis, a disease of the gut mucosa. Advances in recombinant DNA technology and more recently whole genome sequencing of shigellae enabled the development of live-attenuated oral Shigella candidates with defined deletions of virulence factors. These replaced the older attenuated mutants obtained by serial passages in vitro, which showed a certain degree of protection in field studies but required the administration of annual boosters and were associated with a risk of reversion to the virulent phenotype [20,21]. The knowledge on the exact changes in the bacterial genome associated with different levels of attenuation together with the various advantages of the manufacturing and delivery of oral live attenuated vaccines strongly supported the investment of research efforts and funding in the development of this vaccination strategy. This has been the leading approach in Shigella vaccine development at the Walter Reed Army Institute of Research (WRAIR), Center for Vaccine Development (CVD) at University of Maryland, Institut Pasteur, and Karolinska Institute [14,20].
Two major obstacles appeared and significantly slowed down the process of development of these promising candidates. The first was the narrow window between the immunogenicity and safety of these candidates. The development process of the hybrid E. coli K12-S. flexneri 2a (EcSf2a1 and 2), S. flexneri 2a SC602 vaccine strain, S. dysenteriae type 1 (SC599 and WRSd1), the series of S. sonnei WRSS1, WRSS2, and WRSS3, and the series of CVD 1204, CVD 1207, CVD 1208, and CVD 1208S exemplified this delicate balance along the developmental pathway [20].
A second emerging concern was the discrepancy in the potential protective capacity of these live, rationally attenuated oral vaccine candidates when tested in western volunteers and in individuals in endemic areas [22]. For example, the live-attenuated S. flexneri 2a SC602 strain, which was well tolerated and performed excellently in North American volunteers at a single dose of 10 4 CFUs [23] in terms of replication, immunogenicity, and protection, demonstrated very poor excretion and low immunogenicity in Bangladeshi adults and children [23,24]. The immunogenicity data related to the S. flexneri 2a SC602 vaccine strain in Bangladesh corroborated findings on the weaker immunogenicity and efficacy in children in LMICs compared with high income countries experienced with other oral enteric vaccines, including the two licensed live-attenuated rotavirus vaccines [22]. Different hypotheses have been raised to explain the 'intestinal barrier' of volunteers in LMICs against this S. flexneri 2a SC602 live vaccine strain and other oral enteric vaccines and solutions proposed to overcome this obstacle [22,24]. An encouraging example in this direction is the improvement of the immunogenicity of the live-attenuated S. sonnei WRSS1 vaccine strain, which lacks the ability to spread from cell to cell due to the loss of VirG (or IcsA). This was achieved in a recent phase I trial in Bangladeshi adults and children aged 5-9 years after raising the number of oral vaccine doses to three and reducing the volume of bicarbonate buffer to neutralize gastric acidity without increasing reactogenicity [25]. The immune responses elicited by WRSS1 were greater than previously seen with SC602 [24].

Serum IgG antibodies to Shigella LPS Are Linked to Decreased Incidence of Homologous Infection
ELISA was employed in conjunction with passive hemagglutination to prospectively detect anti-LPS antibody rise in individuals involved in epidemics of shigellosis that occurred in Israeli military camps [26,27]. Both tests were sensitive and specific in identifying antibody responses to the corresponding Shigella LPS. The dynamics of the various immunoglobulin classes after the onset of disease exhibited peak levels of IgA at two weeks post symptoms onset, and a decrease to baseline levels within 10 weeks, while serum IgG levels were highest at 3-4 weeks and decreased thereafter. The levels of IgG at late convalescence were half of the levels detected at early convalescence. The IgM titers displayed a dynamic comparable to that of IgA, but the magnitude was lower [27].
The utilization of ELISA or passive hemagglutination assay after the treatment of sera with 2-mercaptoethanol allowed to measure the serum IgG anti-LPS fraction separately by ELISA or the non-IgM fraction by passive hemagglutination, circumventing overexpression of the pentavalent IgM fraction in the passive hemagglutination test. This approach uncovered the importance of these immunoglobulin fractions, and subsequently their association with protection against shigellosis previously masked by the overexpression of IgM in the old agglutination and hemagglutination assays. The pattern of the IgG subclass rise induced by natural Shigella infection was serogroup-dependent. IgG2 was the leading subclass elicited in response to S. flexneri 2a infection while IgG1 and IgG2 were the major fractions induced by S. sonnei infection [28].
Further seroepidemiological studies undertaken among Israeli soldiers highly exposed to Shigella revealed that pre-existent serum IgG antibodies to S. sonnei or to S. flexneri 2a LPS were strongly associated with a reduced risk of homologous Shigella infection [26,29,30]. These studies were carried out in the late 1980s and beginning of the 1990s, when shigellosis was widespread among Israeli recruits serving under field conditions in which Shigella species with a very low infectious dose were easily transmitted feco-orally by multiple routes [7,31]. Under these conditions, soldiers with "low" IgG titers to S. sonnei LPS at baseline had a 5.5-fold (p = 0.0001) higher risk to acquire S. sonnei and develop shigellosis versus soldiers with "high" antibody titers. In agreement with these findings, analysis in S. flexneri 2a outbreaks showed an odds ratio of 4.3 for ELISA serum IgG titers to S. flexneri 2a LPS. These figures correspond to an 82% and 77% reduction in the risk of shigellosis attributed to the homologous serotypes in individuals with high levels of serum IgG antibodies against S. sonnei at baseline.
There was no association between "high" baseline antibody titers to S. sonnei LPS specific antigen and the risk of S. flexneri 2a shigellosis and vice versa [29]. This observation supported the assumption that the protection provided by pre-existing IgG anti-LPS is serotype-specific ( Figure 1). posed to Shigella revealed that pre-existent serum IgG antibodies to S. sonnei or to S. flexneri 2a LPS were strongly associated with a reduced risk of homologous Shigella infection [26,29,30]. These studies were carried out in the late 1980s and beginning of the 1990s, when shigellosis was widespread among Israeli recruits serving under field conditions in which Shigella species with a very low infectious dose were easily transmitted feco-orally by multiple routes [7,31]. Under these conditions, soldiers with "low" IgG titers to S. sonnei LPS at baseline had a 5.5-fold (p = 0.0001) higher risk to acquire S. sonnei and develop shigellosis versus soldiers with "high" antibody titers. In agreement with these findings, analysis in S. flexneri 2a outbreaks showed an odds ratio of 4.3 for ELISA serum IgG titers to S. flexneri 2a LPS. These figures correspond to an 82% and 77% reduction in the risk of shigellosis attributed to the homologous serotypes in individuals with high levels of serum IgG antibodies against S. sonnei at baseline.
There was no association between "high" baseline antibody titers to S. sonnei LPS specific antigen and the risk of S. flexneri 2a shigellosis and vice versa [29]. This observation supported the assumption that the protection provided by pre-existing IgG anti-LPS is serotype-specific ( Figure 1) The incidence of S. sonnei and S. flexneri 2a shigellosis, determined based on visits to the clinic because of diarrhea and a stool culture growing these serotypes, was higher among soldiers with shorter versus longer previous exposure to field conditions (3.3% and 0.05%, respectively, p = 0.05). The percentage of persons with antibodies to S. sonnei or S. flexneri 2a LPS was significantly lower among those who served 0-6 months in the field military units compared to those who served 7-15 months in similar conditions (57.0% and 72.3%, respectively, p = 0.001) [32]. Recurring exposures to Shigella and natural The incidence of S. sonnei and S. flexneri 2a shigellosis, determined based on visits to the clinic because of diarrhea and a stool culture growing these serotypes, was higher among soldiers with shorter versus longer previous exposure to field conditions (3.3% and 0.05%, respectively, p = 0.05). The percentage of persons with antibodies to S. sonnei or S. flexneri 2a LPS was significantly lower among those who served 0-6 months in the field military units compared to those who served 7-15 months in similar conditions (57.0% and 72.3%, respectively, p = 0.001) [32]. Recurring exposures to Shigella and natural boosters seemed to speedily elicit high levels of serum IgG anti-Shigella LPS and further acquired immunity to shigellosis [26,32].
Analogically, the acquisition of IgG anti-LPS antibodies in an age-related mode might be the result of recurring encounters with Shigella LPS and with cross-reacting antigens, even though over a longer time. Similar to capsulated pathogens, such as Haemophilus influenzae type b, Streptococcus pneumoniae, and Neisseria meningitidis, the level of "natural" IgG anti-Shigella LPS is inversely associated with the age-specific incidence rates of shigellosis. The disease incidence is low among infants less than six months of age, is highest at 1-4 years of age, and decreases thereafter [33,34]. Elevated levels of serum IgG antibodies to Shigella LPS were detected in groups of lower socioeconomic status with an increased risk of encounters with Shigella or cross-reacting enteric bacteria earlier in life [30,35]. Moreover, anti-Shigella LPS antibodies increased in an age-dependent pattern among populations living in highly endemic regions and this was linked to a reduced likelihood for shigellosis [36,37].

Construction of Detoxified O-SP-Protein Carrier Shigella Conjugates
The findings of the observational studies in Israel pointing out the role of serum IgG anti-Shigella LPS antibodies in resistance against disease caused by homologous Shigella serotypes, and the development of the groundbreaking effective H. influenzae type b conjugate vaccine, provided the solid basis for Robbins and Schneerson to create injectable glycoconjugates containing O-SP of S. flexneri 2a, S. sonnei, and S. dysenteriae type 1 (Shiga) bound to protein carriers [38,39]. Robbins and Schneerson expected that the accomplishments with capsular polysaccharides could be applied to LPS to construct glycoconjugates consisting of Shigella O-SP, thus obtaining the T-helper cells support to stimulate a strong anti-O-SP T-dependent immune response [38,39].
This approach was novel for a vaccine against shigellosis, a disease of the intestinal mucosa, and was launched while, as described above, a series of live, rationally attenuated oral vaccine strains were extensively investigated.
Robbins and Schneerson postulated that a critical amount of IgG anti-LPS elicited in serum by an injectable conjugate vaccine exudes onto the epithelium of the gut, and in concert with complement, could result in the killing of shigellae reaching the mucosa [34,39].
The endotoxicity of LPS prevents its use as a parenterally administered immunogen. This obstacle was circumvented by the detoxification of LPS to O-SP and binding to a protein carrier. LPS was extracted by the Westphal and Jann hot phenol method [40] and detoxified O-SP was purified by acid hydrolysis [38,41].
The first Shigella conjugate was of O-SP of S. dysenteriae type 1 (Shiga) covalently bound to tetanus toxoid (TT) protein carrier. Conjugates were prepared by three methods with adipic acid dihydrazide (ADH) as a linker and their immunogenicity was further assessed in mice. The first two methods used ADH bound to the carbonyls of ketodeoxyoctonate (KDO) at the reducing end of the core [38,42]. The third method used the random cyanogen bromide (CNBr)-mediated derivatization of O-SP with ADH and consequent conjugation to carboxyl groups of TT in which O-SP and TT are linked at multiple sites [38,42]. The antibody response in serum samples of mice injected three times, two weeks apart, was the highest after immunization with the conjugates prepared with multipoint attachments using ADH as a spacer ("lattice"-type conjugates) following each injection [38,42]. Subsequent assessment of adjuvanted formulations of the latter revealed that the use of both alum and monophosphoryl lipid A (MPL) in an oil/water emulsion enhanced the induced anti-LPS antibody response [38,42]. The conjugation approach used with S. dysenteriae type I was further employed for the construction and evaluation of S. flexneri 2a and S. sonnei conjugates [41].
In an effort to reach the highest immunogenicity of the S. flexneri 2a LPS-based conjugates, random CNBr-mediated derivatization of DeLPS (LPS detoxified with hydrazine) and O-SP (LPS detoxified with acetic acid) with ADH as linker were conjugated to carboxyl groups of TT and compared [43]. The DeLPS material was of greater molecular weight versus the analogous O-SP but the saccharide:protein molar ratio was higher (3:1 to 5:1 versus 2:1) for conjugates including a O-SP component. Both conjugate formulations induced anti-LPS serum IgG antibodies in mice, but the S. flexneri 2a O-SP was a superior T-celldependent immunogen through its covalent coupling to TT [43]. The LPS acid-mediated detoxification process was adopted in subsequent developments.
Carriers other than tetanus toxoid were further investigated. S. flexneri 2a O-SP conjugates were constructed by means of random conjugation chemistry, utilizing a nontoxic recombinant mutant of the Pseudomonas aeruginosa exoprotein A (rEPA). rEPA was selected since the presence of antibodies to this carrier is much less expected in humans in comparison to antibodies to TT or diphtheria toxoid induced by routine national vaccination programs. This carrier can also have medical importance, inducing protective antibodies against opportunistic invasive infections caused by Pseudomonas aeruginosa [44]. These conjugates induced anti-LPS IgG antibodies in mice [41]. Moreover, whereas the alum adjuvanted and non-adjuvanted conjugates induced similar anti-LPS IgM antibody levels, the alum adsorbed conjugate elicited a higher level of anti-S. flexneri 2a LPS IgG in mice after a third injection [41].
Similarly, the S. sonnei O-SP-protein conjugates used random chemical conjugation, ADH as a spacer, and rEPA as the protein carrier. However, whereas hydroxyl groups were involved as ADH anchoring sites in the case of S. dysenteriae type 1 and S. flexneri 2a O-SP, ADH was directly linked to the carboxyl groups of the S. sonnei O-SP LPS. Immunogenicity analysis revealed a trend similar to that of the S. flexneri 2a conjugates [41]. The conjugation process was modified in an attempt at increasing conjugate solubility, limiting aggregation, and avoiding extensive intramolecular and intermolecular cross-linking. It was found that succinylation of the amino group of the protein carrier increased the conjugation yield and improved the immunogenicity of the conjugates in young, outbred mice [42,45].

Assessment of the Safety, Immunogenicity and Protective Efficacy of the Shigella Conjugate Vaccines in Adults
In a phase I clinical trial conducted in the US, adult volunteers were administered two doses of one of three conjugates, S. dysenteriae type 1 O-SP-TT, S. flexneri 2a O-SP-rEPA and S. sonnei O-SP-rEPA at 25µg of O-SP and 75µg protein carrier [41]. The conjugates (0.5 mL) were injected intramuscularly (IM), 6 weeks apart, with or without alum as adjuvant. They demonstrated a good safety profile and were immunogenic. S. flexneri 2a-rEPA and S. sonnei-rEPA conjugates induced serum IgG and IgM responses similar or higher than those of young Israeli adults recovering from shigellosis. Unlike the mouse data, there was no enhancement of the IgG and IgM response after the second injection, and no increased antibody response was observed in recipients of the conjugates given with alum [41].
In a phase II vaccine-controlled clinical trial that assessed the safety and immunogenicity in young adults in Israel, 66 volunteers were vaccinated with S. sonnei-rEPA vaccine and 64 received the S. flexneri-rEPA vaccine. Of those, 17 and 16 participants, respectively, received a second injection of the same conjugate six weeks following vaccination with the first dose. Sixty two volunteers received a hepatitis B vaccine as control vaccine [46]. Fourteen days post vaccination with the first dose, 90% of S. sonnei-rEPA vaccinees and 73-77% of S. flexneri-rEPA vaccinees had a four-fold or greater rise in serum IgG and IgA anti-LPS levels, while none of the hepatitis B vaccine recipients presented a significant antibody increase to either LPS. The second dose given on day 42 did not boost antibody titers [46], as was also reported in the phase I trial in the US. Four years after vaccination, 50% of recipients of Shigella vaccines still showed four-fold or higher titers, relative to baseline IgG titer. Serum IgG antibody level was the highest and most persistent class of LPS antibodies (Figure 2).
The increase in serum IgG subclasses in sera of volunteers who were vaccinat the S. flexneri 2a-rEPA or S. sonnei -rEPA vaccine followed a similar pattern to that i by the corresponding Shigella infection. IgG2 was the major subclass induced in re to S. flexneri 2a O-SP, whereas IgG1 and IgG2 were the main IgG subclasses elicite response to S. sonnei LPS O-SP [28,47]. rEPA was the carrier linked to both the O-S sonnei and S. flexneri 2a. Therefore, we assume that the different pattern of IgG a subclass response is associated with the chemical structure of the two O-SPs [47].
Interestingly, a significant anti-LPS IgA and IgG response in urine was also ob in 60% of the recipients of the S. sonnei-rEPA conjugate [48]. The elicited IgA in ur of secretory origin and there was a high correlation between the urinary and seru body responses [48]. These data suggested that, in addition to a strong serum sponse, S. sonnei-rEPA also induces general mucosal stimulation expressed in the urinary secretory IgA.
A double-blind randomized vaccine-controlled phase III clinical trial evalua protective efficacy of a single dose of the S. sonnei-rEPA vaccine among 1446 rec seven different cohorts (field sites) in Israel at high risk of exposure to Shigella [4 unteers were randomly allocated to three vaccine groups and received one injecti The increase in serum IgG subclasses in sera of volunteers who were vaccinated with the S. flexneri 2a-rEPA or S. sonnei -rEPA vaccine followed a similar pattern to that induced by the corresponding Shigella infection. IgG2 was the major subclass induced in response to S. flexneri 2a O-SP, whereas IgG1 and IgG2 were the main IgG subclasses elicited in the response to S. sonnei LPS O-SP [28,47]. rEPA was the carrier linked to both the O-SPs of S. sonnei and S. flexneri 2a. Therefore, we assume that the different pattern of IgG anti-LPS subclass response is associated with the chemical structure of the two O-SPs [47].
Interestingly, a significant anti-LPS IgA and IgG response in urine was also observed in 60% of the recipients of the S. sonnei-rEPA conjugate [48]. The elicited IgA in urine was of secretory origin and there was a high correlation between the urinary and serum antibody responses [48]. These data suggested that, in addition to a strong serum IgG response, S. sonnei-rEPA also induces general mucosal stimulation expressed in the rise of urinary secretory IgA.
A double-blind randomized vaccine-controlled phase III clinical trial evaluated the protective efficacy of a single dose of the S. sonnei-rEPA vaccine among 1446 recruits in seven different cohorts (field sites) in Israel at high risk of exposure to Shigella [49]. Volunteers were randomly allocated to three vaccine groups and received one injection of S sonnei-rEPA and four doses of oral placebo; four oral doses of the hybrid Escherichia coli K12-S. flexneri 2a (EcSf2a-2) vaccine and one injection of saline (placebo); or one injection of meningococcal control vaccine and four doses of oral placebo. During the follow-up after vaccination, culture-proven cases of S. sonnei shigellosis occurred in three cohorts, 70-155 days post-vaccination and in one cohort, as early as 1-17 days post-vaccination. In the first three cohorts, the incidence of shigellosis was 2.2% among vaccinees who received the S sonnei-rEPA vaccine versus 8.6% in the control vaccines groups (p = 0.006), yielding a protective efficacy of 74% (95% CI 28-100). Moreover, S sonnei-rEPA conferred significant protection against culture-proven S. sonnei shigellosis in the cohort in which cases occurred 1-17 days post-vaccination, yielding 43% vaccine efficacy (95% CI 4-82), p = 0.039 [49]. There were no cases of S. flexneri 2a shigellosis among the volunteers during the active surveillance periods. Thus, the efficacy of the EcSf2a-2 vaccine could not be assessed. Pre-vaccination levels of serum IgG and IgA antibodies to S. sonnei LPS were comparable in recipients of S. sonnei-rEPA vaccine and the control vaccines. S. sonnei-rEPA vaccinees who developed S. sonnei shigellosis had significantly lower serum IgG responses to the homologous LPS than vaccinees who did not develop S. sonnei shigellosis. The higher serum IgG antibody response elicited by the S. sonnei-rEPA conjugate vaccine in the group that did not develop shigellosis (p = 0.014) confirmed the previous observations concerning the relationship between IgG serum antibody level and protection against homologous disease [26,49] (Figure 3). occurred 1-17 days post-vaccination, yielding 43% vaccine efficacy (95% CI 4-82), p = 0. [49]. There were no cases of S. flexneri 2a shigellosis among the volunteers during the tive surveillance periods. Thus, the efficacy of the EcSf2a-2 vaccine could not be assess Pre-vaccination levels of serum IgG and IgA antibodies to S. sonnei LPS were compara in recipients of S. sonnei-rEPA vaccine and the control vaccines. S. sonnei-rEPA vaccin who developed S sonnei shigellosis had significantly lower serum IgG responses to homologous LPS than vaccinees who did not develop S. sonnei shigellosis. The higher rum IgG antibody response elicited by the S. sonnei-rEPA conjugate vaccine in the gro that did not develop shigellosis (p = 0.014) confirmed the previous observations conce ing the relationship between IgG serum antibody level and protection against homo gous disease [26,49] (Figure 3).

Assessment of the Safety, Immunogenicity and Protective Efficacy of the Shigella Conjugate Vaccines in Children
In view of the favorable results of the immunogenicity and protective efficacy of the S. sonnei-rEPA conjugate vaccine among young adults, age-descending phase II randomized controlled trials were performed with S. sonnei and S. flexneri 2a conjugates in Israeli healthy children 4-7 years old [50] and 1-4 years old [51]. Results of these studies showed a very good safety profile of the conjugate vaccines in children, confirming safety data previously generated in young adults.
In the age group 4-7 years, both S. sonnei and S. flexneri 2a O-SP-rEPA conjugates (at 25 µg saccharide and 75 µg protein concentrations), non-adjuvanted, induced significant responses in serum IgM, IgA and IgG to homologous LPS. The highest and most sustained response was in IgG with a four-fold rise in 96% and 98% of vaccinees who received one dose of the S. sonnei or S. flexneri 2a conjugate, respectively [50]. The second injection of the conjugates induced a homologous booster rise in children given the S. flexneri 2a-rEPA but not S. sonnei -rEPA vaccines. The anti-LPS IgG antibody titers measured at six months after the second dose declined but were still significantly higher than the pre-vaccination levels.
In the phase II study in children aged 1-4 years, the Shigella conjugates evaluated were S. sonnei-O-SP linked to CRM9 and S. flexneri 2a O-SP linked to succinylated rEPA. The S. flexneri 2a-O-SP rEPA succ conjugate was more immunogenic compared to the respective conjugates containing rEPA and CRM9 as carrier proteins in studies in mice and subsequently in adult volunteers [52]. In contrast, succinylation did not improve the im-munogenicity of the S. sonnei O-SP when conjugated to either CRM9 or rEPA [52]. S. flexneri 2a-O-SP rEPAsucc conjugate was further used in young children.
The S. sonnei -CRM9 and S. flexneri 2a-rEPA succ elicited a four-fold or greater rise in titer in 92% and 85% of volunteers, after the second injection, respectively. The S. sonnei conjugate induced a booster response whereas the rise in anti-LPS IgG antibody level elicited by the second dose of S. flexneri 2a conjugate was not statistically significant. Two years later, the GMT of IgG anti-LPS induced by both conjugate vaccines were still significantly higher compared to pre-vaccination.
Pre-vaccination and post-vaccination levels of IgG anti-LPS differed according to age, being lower in children compared to adults. The geometric mean levels of homologous IgG anti-LPS at four weeks following vaccination with the second dose of the S. sonnei conjugate vaccine were 48.0, 8.0, and 2.9 in adults, children aged 4-7 years and 1-4 years, respectively. The corresponding levels for S. flexneri 2a IgG anti-LPS were 113.0, 48.0, and 40.1 [50][51][52]. In all these trials, no heterologous anti-S. sonnei or S. flexneri 2a LPS rises were found, confirming the possibility of evaluating the two conjugates in a single trial and having them play the role of control vaccine, each one for the other. All studies also documented a satisfactory response to the carrier proteins used. There is no clear-cut explanation for the variability seen in booster effect of the second dose of each of the S. sonnei and S. flexneri 2a conjugates in the clinical studies. The variation observed could have been vaccine construct and vaccinee (host)-related, or the result of an interaction between both factors.
In view of these data, a double-blind, randomized and vaccine-controlled efficacy trial of S. sonnei-rEPA and S. flexneri 2a O-SP-rEPA succ conjugates (non-adjuvanted) was carried out among 2799 healthy children aged 1-4 years in Israel [53]. Two intramuscular injections of the conjugates were administered to children six weeks apart. Serum samples were obtained from randomly selected 10% of the children and IgG anti-LPS levels were measured by ELISA. The number of cases of S. flexneri 2a shigellosis was very small, which prevented an efficacy analysis of S. flexneri 2a O-SP-rEPA succ in this study. The pooled efficacy of the S. sonnei conjugate in the prevention of culture-documented S. sonnei shigellosis was 27.5%. Nonetheless, an age-stratified analysis revealed an age-dependent efficacy for children who were vaccinated with the S. sonnei conjugate vaccine: 3.8%, 35.5%, and 71.1% (p = 0.043) for children aged 1-2 years, >2-3 years, and >3-4 years, respectively. Likewise, an age-dependent rise was evident in serum antibody levels elicited by the S. sonnei vaccine. The levels of IgG antibodies in ELISA units (EU) against-LPS of S. sonnei were 1.4, 3.71, and 6.38 in the age groups 1-2, >2-3, and >3-4 years, respectively [53]. The efficacy figures paralleled the age-related immunogenicity of the S. sonnei conjugate. These "dose-response trend" associations suggest that a defined level of serum IgG anti-O-SP antibodies induces protection against shigellosis [26]. Namely, the threshold level associated with 71% protection in the age group 3-4 years in this trial can provide a reference value when evaluating the immunogenicity of the S. sonnei component in the new generation of tetravalent conjugates in infants and toddlers, and for forecasting their efficacy in the pediatric target population. At the age groups and surrogate baseline values deduced from children receiving the S. flexneri 2a-rEPA control vaccine, since no paired blood samples were obtained from children in this study, there was no correlation between the fold increase in antibody level of the three age groups and the efficacy of the S. sonnei conjugate. Such correlations may occur in older age groups or among adults, where baseline values would be higher and more consistent.
The main findings on the immunogenicity and efficacy of the various Shigella conjugate constructs are summarized in Table 1. Table 1. Shigella conjugate constructs and immunogenicity and efficacy in age descending studies.

Article Stage in Clinical Development Vaccine Constructs and Study Groups Main Findings
Taylor

Clinical Evaluation of the Detoxified O-SP-rEPA Shigella Conjugates Confirmed That IgG Antibodies to Shigella LPS Are Correlates of Protection
An important contribution of the detoxified O-SP-rEPA Shigella conjugates clinical evaluation has been the confirmation that IgG antibodies to Shigella LPS are correlates of protection with mechanistic capabilities.
Adapting prior definitions and criteria for correlates of protection [54] to shigellosis, we found that serum IgG antibodies to Shigella LPS meet all the criteria and can be defined as a correlate of protection [26]. Shigella anti-LPS serum IgG antibodies are induced by Shigella infection and they have been linked with resistance to shigellosis after natural exposure. Their titers increase in an age-dependent mode in parallel with a significant decrease in the incidence of shigellosis, and they have demonstrated a dose-response association with the protection level achieved by the S. sonnei conjugate evaluated in efficacy studies in young adults and children. These collective data have been reemphasized by findings of a recent controlled human infection model (CHIM) study showing that LPS-specific serum IgG responses in bioconjugate vaccine (Flexyn2a) recipients were associated with protection against S. flexneri 2a homologous disease [55].
Serum IgG antibodies to Shigella LPS also exhibited mechanistic protective capabilities. Using the serum bactericidal assay (SBA) [56,57] and utilizing a stable S. sonnei phase 1 strain as a target for SBA, we identified a significant increase in SBA titers in the sera of S. sonnei-rEPA vaccinees (GMT = 1472; 95% CI: 690-3143, n = 25) three months following vaccination, in comparison to a GMT of 50 (95% CI: 31-80) before vaccination [26]. The titers correlated with the IgG anti-S. sonnei LPS titers.
Employing the thiocyanate elution assay to measure avidity [58], we found that sera bactericidal activity had a superior avidity index versus sera with no bactericidal activity on S. sonnei (Avidity index: 2.3 vs. 1.8, p = 0.048) [26]. Post-but not pre-vaccination sera of children receiving the S. sonnei or S. flexneri 2a O-SP conjugates in the frame of the age-descending immunogenicity trials blocked the in vitro invasion of Caco-2 cells and inhibited rises in IL-1β and IL-8 mRNA and extracellular cytokine levels [59].
We are currently working to identify thresholds of IgG anti-S. sonnei LPS that predicted specific levels of vaccine efficacy in the phase III study in young adults and in the subsequent efficacy trial in children in Israel. Altogether, the data on IgG anti-S. sonnei LPS as a correlate of protection stemming from the evaluation of the detoxified O-SP-rEPA Shigella conjugates can support and guide the clinical development of the new generation of conjugates, namely bioconjugates and synthetic carbohydrate-based conjugates, or the subunit generalized modules for membrane antigen (GMMA)-based vaccines [55,[60][61][62]. It would be of much value to also corroborate these data on serum IgG anti-S. sonnei LPS correlate of protection and proposed thresholds in observational studies and clinical trials of the vaccine candidates in LMIC children who are the target for future immunization. We cannot rule out the possibility that a host who has previously experienced mucosal exposure to Shigella antigens (e.g., children above three years of age) may be better protected by the parenteral administration of a conjugate vaccine than infants and very young children. Recent studies that we carried out on consecutive plasma samples collected from Zambian infants at the age of 6, 14 and 52 weeks are consistent with early exposure to Shigella and indicate naturally acquired IgG and IgA antibodies to S. flexneri 2a and S. sonnei LPS in a proportion of infants between 14 and 52 weeks of age [63]. At the age of nine months, targeted for the first dose of a Shigella vaccine in such settings, the oral natural priming documented by the early rise in anti-LPS antibodies may play a supportive role in the immune response and subsequent protection following the parenteral delivery of immunogenic vaccine candidates.

O-SP Acetylation and Potential Cross-Reactivity/Protection between S. flexneri Serotypes
Though there were not enough cases of S. flexneri 2a shigellosis to the assess efficacy of the S. flexneri 2a-rEPA conjugate against homologous S. flexneri 2a shigellosis in the efficacy trial conducted among children aged 1-4-years, the S. flexneri 2a conjugate showed a trend of protection (~50%, albeit not reaching statistical significance) against S. flexneri serotype 6 infection in the age group 3-4 years [53]. It was hypothesized that the O-acetylated disaccharide rhamnose present on the same position in each one of the two O-SPs backbones, which would otherwise have a different structure, could account for the observed crossreactivity between S. flexneri 2a and S. flexneri 6 LPSs. (Figure 4). A more recent study utilizing residual serum samples from S. flexneri 2a conjugate vaccinees involved in previous clinical trials demonstrated that 20-32% of adults and children 3-7 years of age receiving one dose or two doses of conjugate, respectively, had a ≥4-fold rise in anti-S. flexneri 6 LPS and showed no rise in antibodies to S. sonnei LPS, which served as negative control [64]. These and additional data from inhibition studies with homologous and heterologous soluble O-SPs in humans were in disagreement to the findings in mice and guinea pigs that reported a complete absence of cross-reactivity between the two Shigella LPSs [15,64]. These original findings have been the basis for studies underway at Tel Aviv University on the potential cross-reactivity between S. flexneri 2a and S. flexneri 6 LPSs by means of binding and functional antibodies, maintaining a two-side hypothesis concerning the role of acetylation of the O-SP in the vaccines and/or in the in the coating LPS. Significant cross-reactivity and subsequent potential cross-protection between these most common S. flexneri serotypes could perhaps allow a reduction in the number of costly multivalent vaccine components.

O-SP Acetylation and Potential Cross-Reactivity/Protection between S. flexneri Serotypes
Though there were not enough cases of S. flexneri 2a shigellosis to the assess efficacy of the S. flexneri 2a-rEPA conjugate against homologous S. flexneri 2a shigellosis in the efficacy trial conducted among children aged 1-4-years, the S. flexneri 2a conjugate showed a trend of protection (~50%, albeit not reaching statistical significance) against S. flexneri serotype 6 infection in the age group 3-4 years [53]. It was hypothesized that the O-acetylated disaccharide rhamnose present on the same position in each one of the two O-SPs backbones, which would otherwise have a different structure, could account for the observed cross-reactivity between S. flexneri 2a and S. flexneri 6 LPSs. (Figure 4). A more recent study utilizing residual serum samples from S. flexneri 2a conjugate vaccinees involved in previous clinical trials demonstrated that 20-32% of adults and children 3-7 years of age receiving one dose or two doses of conjugate, respectively, had a ≥4-fold rise in anti-S. flexneri 6 LPS and showed no rise in antibodies to S. sonnei LPS, which served as negative control [64]. These and additional data from inhibition studies with homologous and heterologous soluble O-SPs in humans were in disagreement to the findings in mice and guinea pigs that reported a complete absence of cross-reactivity between the two Shigella LPSs [15,64]. These original findings have been the basis for studies underway at Tel Aviv University on the potential cross-reactivity between S. flexneri 2a and S. flexneri 6 LPSs by means of binding and functional antibodies, maintaining a two-side hypothesis concerning the role of acetylation of the O-SP in the vaccines and/or in the in the coating LPS. Significant cross-reactivity and subsequent potential cross-protection between these most common S. flexneri serotypes could perhaps allow a reduction in the number of costly multivalent vaccine components.

Efforts to Further Enhance the IgG Response to Shigella O-SP-Protein Conjugates
In search for a more immunogenic conjugate that can elicit a critical IgG anti-LPS level to confer protection in infants and young children, it was found that protein conjugates of synthetic saccharides could induce higher levels of serum IgG LPS antibodies in mice than those of the O-SP from S. dysenteriae type 1 [65]. The O-SP of S. dysenteriae type 1 (approximately 27 tetrasaccharide repeat units), prepared by acid hydrolysis of the LPS, was bound to human serum albumin (HSA) by multiple point attachment (O-SP-HSA): The molar ratio of HSA to O-SP was 1.0. Synthetic saccharides, composed of one or multiples of the O-SP tetrasaccharide, equipped with a spacer at their reducing end, were bound to HSA by a single point attachment: The average molar ratios of the saccharides to HSA ranged from 4 to 24. The conjugates of the octamer, dodecamer, and hexadecamer synthetic saccharides elicited IgG LPS antibodies after the second injection, a statistically significant rise (booster) after the third injection, and higher levels than those vaccinated with O-SP-HSA (p = 0.0001). The highest geometric mean levels of IgG anti-LPS were elicited by the hexadecamer with nine chains or 9 moles of saccharide/HSA (15.5 ELISA units), followed by the octamer with 20 chains (11.1 ELISA units) and the dodecamer with 10 chains (9.52 ELISA units) [65].
This breakthrough finding in mice on the immunogenicity of protein conjugates of the S. dysenteriae type 1 synthetic saccharides was a landmark for the development of the novel Shigella synthetic carbohydrate-based conjugates at Institut Pasteur [42,66]. S. flexneri 2a-TT conjugate (Sf2a-TT15) original synthetic oligosaccharide-protein conjugate administered at two concentrations, alum-adjuvanted and non-adjuvanted, was safe and well tolerated and induced high titers of anti-S. flexneri 2a LPS IgG antibodies in a firstin-human phase I study conducted in Israel [60]. The immune response induced by the Sf2a-TT15 first carbohydrate-based synthetic conjugate evaluated in humans was stronger than that elicited by the detoxified S. flexneri 2a-rEPA evaluated in a similar young adult population [25].

Conclusions
The classical detoxified glycoconjugates have changed the paradigm of Shigella vaccine development, demonstrating that a parenterally delivered potent vaccine can protect against a mucosal intestinal infection. All stages of the preclinical and clinical development of the classical conjugates provided valuable data that have triggered and guided the development of the current new generation of mono and tetravalent glycoconjugates and other injectable O-SP based Shigella vaccine candidates. It is assumed that the lessons learned, the contribution of the new methodological advances, and the current financial support will lead to the licensure of the desired vaccine(s) that will fill the gap and confer protection against shigellosis to infants and very young children in highly endemic regions all over the world.  Acknowledgments: This article is dedicated to the late John B. Robbins and to Rachel Schneerson (long life to her) for their visionary approach in developing the first generation of injectable Shigella glycoconjugate vaccines. The authors wish to thank all researchers and collaborators in Israel and in the US for their major contribution to the development of Shigella conjugate vaccines from observational studies in the field to proof of concept in phase III efficacy trials as presented in this review.