Glycoconjugate vaccines: Principles and mechanisms

See allHide authors and affiliations

Science Translational Medicine  29 Aug 2018:
Vol. 10, Issue 456, eaat4615
DOI: 10.1126/scitranslmed.aat4615


Bacterial conjugate vaccines are used in infants, adolescents, and the elderly, and they are among the safest and most successful vaccines developed during the last 40 years. Conjugation of polysaccharides to proteins provides T cell epitopes that are necessary in the germinal centers for the affinity maturation of polysaccharide-specific B cells. Collective analysis of data from animal experiments and clinical trials, reviewed with current knowledge of immunology, revealed possible mechanistic explanations that may improve our understanding of conjugate vaccines. Key conclusions are that naïve infants respond differently from adolescents and adults and that most of recommended schedules generate only 10 to 35% of the maximal antibody titer that the vaccine can induce, indicating that the full potential of glycoconjugate vaccines has not yet been reached.


Conjugate vaccines are composed by covalently linking a bacterial polysaccharide to a protein. Building on seminal research in the early decades of the last century (1, 2), conjugate vaccines were developed starting in the mid-1970s after observations that the capsular polysaccharides of Haemophilus influenzae type b (Hib) (3) and meningococcus C (4) failed to induce protective antibodies when used as vaccines in young children. Now, we know that most polysaccharides are not good vaccines because, on their own, they are unable to interact with the receptors on T cells in the germinal centers (GCs). However, through peptides derived from the processing of protein covalently linked to polysaccharide, T cells are engaged and stimulated (5). The conjugate vaccines against bacteria that have been developed during the last 40 years are one of the success stories of modern vaccination. Today, we have successful vaccines licensed worldwide against Haemophilus influenzae; meningococcus serogroups A, C, and ACWY; 10 to 13 serotypes of pneumococcus; and Salmonella typhi. Together, these vaccines have had a huge impact on global infant mortality and morbidity by eliminating some of the historical scourges of mankind. For example, meningococcus C was eliminated from the United Kingdom after a massive vaccination campaign in 1999 (6), and outbreaks of meningococcus A invasive disease have been eliminated from the African meningitis belt (7). Overall, there have been marked reductions in the global occurrence of bacterial meningitis and pneumonia.

During the development and deployment of these vaccines, many polysaccharides conjugated to different carrier proteins were tested in several hundred preclinical studies in animal models and in hundreds of clinical trials, often in different age groups. The data, generated in clinical trials conducted over many years, often following the same subjects over time, were published decades apart in different papers and journals, confirming the excellent immunogenicity and huge impact on disease by these vaccines. However, there has been little progress on the understanding of the behavior of conjugate vaccines in different populations and immunization schedules. This Perspective tries to distill the key features that are common to the preclinical and clinical data, with the aim to gather a better understanding of the behavior of conjugate vaccines in the clinic and in animal models and to link the observed behavior to the mechanistic explanation. This should help to better understand how conjugate vaccines work and possibly learn how to fully exploit their potential. The focus is on what we believe is the most important feature of conjugate vaccines, which is the induction of polysaccharide-specific memory B cells. We will not focus on the T carrier–specific memory T cells because, while they are essential for the immune response to conjugates by providing help to B cells, their help is not specific for the desired immune response and is interchangeable. Children primed with a conjugate vaccine containing one carrier protein can be boosted with the same polysaccharide conjugated to a different protein and vice versa (8).

Age-dependent immune responses to glycoconjugate vaccines

The immune response to conjugate vaccines is summarized in Fig. 1 and varies depending on age. In infants, antibodies are barely detectable after the first dose, and they achieve significant protective titers after the second dose. The third dose of the primary immunization series usually increases the immune response only slightly. The antibody level then decreases over time but usually remains above that of nonimmunized children. If a booster immunization is given 3 to 5 years later, then the antibody levels reach a titer that usually is 3 to 10 times the maximal titer achieved by the primary immunization, indicating that the primary immunization induced memory B cells. In marked contrast, the unconjugated polysaccharide is not able to induce any response in this age group (Fig. 1A). There are a couple of exceptions to the above observations. The first one is that the meningococcus A–unconjugated polysaccharide seems to be immunogenic and to induce some memory in infants (9); the second is that a highly efficacious Hib vaccine, where the polysaccharide was conjugated to outer membrane vesicles of meningococcus B, induced high titers of antibodies after the first dose, which could not be boosted by subsequent immunizations (10, 11). Currently, we do not understand why these vaccines behave differently.

Fig. 1 Antibody titers induced by conjugate vaccines in different age groups.

The absolute titers achieved in infants, toddlers, and adolescents are different and often not comparable because these were performed at different times, often with different methods. Therefore, the percentage of maximal response in three panels does not reflect a similar absolute titer. (A) Data for infant vaccination were derived from the bactericidal titers against meningococcus C reported by Fairley et al. (19) and MacLennan et al. (20). The two primary vaccination trials were conducted in different populations using the same lot of meningococcus AC-CRM conjugate vaccine. The booster immunization in (18) was given using a polysaccharide vaccine. Data are reported as the percentage of the maximal response achieved in these trials. Similar behavior has been reported by Scheifele et al. (21, 22) for Hib-tetanus toxoid conjugate vaccine and by Trück et al. (23) for pneumococcal conjugates. (B) Data from toddlers were derived from the bactericidal titers reported by Halperin et al. (24) and Block et al. (25) who used a meningococcal ACWY-CRM conjugate vaccine. Data are reported as the percentage of the maximal response achieved in these trials, and each point is the average response to serotypes A, C, W, and Y. (C) Data for adolescents were derived from Jackson et al. (26), Baxter et al. (27), and Gill et al. (28) from trials performed using a meningococcal ACWY-CRM conjugate vaccine. Data are reported as the percentage of the maximal response achieved in these trials, and each point is the average response to serotypes A, C, W, and Y. Data for the relative response of polysaccharide vaccines were derived from Anderson et al. (29) and Black et al. (30).


In toddlers, the immune response to conjugate vaccines is similar to infants (Fig. 1B). The response is barely detectable after the first dose, reaches protective titers after two doses, and then decreases over time but remains higher than that of nonimmunized toddlers. If a booster dose is given after three or more years (to those previously immunized with the conjugate and to a control group that had not been vaccinated), then the antibody levels in the primed cohort reach 3 to 10 times the levels induced by the primary immunization, whereas the naïve cohort (control group) responds with very low titers. Because this is also observed when the boost is given with heterologous carrier, this result proves that the primary immunization generated memory B cells that are now able to respond quickly by differentiating into plasma cells to secrete antibodies.

In adolescents, the immune response is quite different (Fig. 1C). The protective antibody titers are not only achieved and peak after the first immunization and decrease over time but also remain higher than those of nonvaccinated adolescents. If a booster dose is given three or more years after the first dose, then the antibody titers increase to levels that are 5 to 10 times higher than those induced by the first immunization. This confirms that one dose of vaccine in adolescents is able to stimulate memory B cells that are later able to respond quickly to a booster immunization. In this age group, the polysaccharide vaccine is also able to induce an antibody response. The response to the polysaccharide is similar or slightly lower than that of the conjugate and peaks after the first dose. However, if a booster dose is given, then there is not a booster response and the antibody response is usually lower than that of the primary immunization. This indicates that, in adolescents, the polysaccharide is not able to induce memory B cells. There is evidence that immunization with the polysaccharide induces apoptosis of memory B cells and reduces the response to subsequent immunizations, a phenomenon known as hyporesponsiveness (12).

Vaccine adjuvants composed of oil in water emulsions such as MF59 or containing Toll-like receptor (TLR) agonists such as AS04 or alum TLR7 behave differently in different age groups. Figure 2A shows that, in animal models (infant baboons or mice), MF59 or AS04 induces a much better immune response than unadjuvanted conjugates or conjugates adjuvanted with alum alone. Although clinical data in infants with such adjuvants are not available, it is reasonable to predict that similar findings would occur in infants and that the presence of adjuvants during priming would result in an increased pool of memory B cells. In marked contrast, in adolescents and adults, the addition of adjuvants, such as MF59 or AS04, does not increase the response to the one-dose immunization (Fig. 2B). In adolescents, two or three doses of conjugates, each spaced 30 to 60 days apart, do not improve the response to conjugate vaccines, regardless of whether they are given with adjuvants or alone.

Fig. 2 Effects of adjuvants and multiple doses on the immunogenicity of conjugate vaccines in animal models and in adolescents.

(A) Data for the use of adjuvants in animal models were derived from Granoff et al. (31), a study in infant baboons using meningococcus C-CRM conjugate vaccine and MF59 as adjuvant. Data are reported as the percentage of the maximal response achieved. The study has been selected as representative of many preclinical studies performed with different conjugates, different adjuvants, and different animal models. Another typical example is reported in (32). (B) Data were derived as follows: The response to a second immunization was derived from Costantino et al. (33) and confirmed by many other studies using different conjugates and different carrier proteins. The data on the inability of adjuvants to increase the response to conjugate vaccines in adolescents are from Leroux-Roels et al. (34). The data about the booster 3 years after the first immunization derive from the clinical trials reported in Fig. 1C.


Immune response in naïve or primed populations

To interpret the clinical and preclinical data reported above, the mechanisms of how the immune response is generated need to be understood. Figure 3A illustrates the conventional mechanism of an immune response in naïve infants or animals. The conjugate vaccine is taken up by dendritic cells at the site of immunization and, within a few days, is transported to lymph nodes where the presence of a T cell–dependent antigen triggers the formation of GCs. GCs are sites within lymph nodes and the spleen where mature B cells proliferate, differentiate, and mutate their antibody genes through somatic hypermutation. This achieves higher affinity and switching of the class of antibodies [for example, from immunoglobulin M (IgM) to IgG] during a normal immune response to an infection or after vaccination. The dynamic changes in GCs occur in spatially distinct regions of the GC, called the light and the dark zone (13, 14). GCs contain antigen-specific B cells expressing the antibody on their surface as a receptor (BCR), T follicular helper (TFH) cells, and follicular dendritic cells (FDCs) that contain and present the antigen to the B cells. B cells in the GC undergo apoptosis unless they are positively selected by interacting with TFH cells and antigen. Selection occurs in the light zone. B cells bind and extract the antigens of the conjugate from the FDCs and process and present the protein-derived antigen to the TFH cells, which provide help to the B cell by direct cell-cell interaction and by secreting cytokines. The activated B cells at this point enter the dark zone, where they multiply rapidly and express activation-induced cytidine deaminase, which triggers the introduction of random mutations in the Ig gene encoding the variable region of the BCR by mutating cytosine to uracil, thereby creating a mismatch in the DNA. Although many of the mutated B cells in the dark zone will undergo apoptosis because of mutations that damage the BCR (for example, introduction of stop codons or mutations that prevent proper protein folding and surface expression) or create self-reactivity, B cells with functional receptors reenter the light zone, where they retrieve antigen from the surface of FDC, which they process and present to TFH cells. Competition among B cells for limiting numbers of TFH cells ensures that B cells bearing affinity-enhancing mutations, which present more antigen to TFH cells and receive more help, are preferentially selected to reenter the dark zone and start the process of affinity maturation again. Over time, a fraction of selected B cells exit the GC to become memory B cells and plasma cells. The GC reaction can be potentiated by the use of adjuvants such as MF59 or TLR agonists, resulting in higher titers of antibodies and an increased number of memory B cells.

Fig. 3 Proposed mechanism for the immune response to conjugates and to polysaccharides in infants and in adolescents and adults.

(A) Conjugates in infants, (B) polysaccharides (PS) in infants, (C) conjugates in adolescents, and (D) polysaccharides in adolescents.


Whereas conjugate vaccines contain B and T cell epitopes and can induce normal affinity maturation of B cells in GC, polysaccharide vaccines (Fig. 3B) lack T cell epitopes. After immunization of infants or naïve animals with polysaccharides, the specific B cells cannot get the help from T cells, and therefore, GC fails to form. The result is a complete absence of a T cell–dependent immune response and ensuing affinity maturation by B cells.

For glycoconjugate vaccines, the situation in adolescents, adults, and the elderly is different. A reasonable explanation is that, with few exceptions, these individuals have acquired preexisting memory B cells specific for the polysaccharide. Although the mechanism(s) by which these memory B cells are generated is unclear, one currently favored hypothesis is that it is the outcome of exposure to the relevant pathogen or to cross-reacting polysaccharides produced by the commensal flora or food. Because thousands of different polysaccharides produced by bacteria are all based on a few common sugars joined by different sequences and different linkages, it is possible that each individual person may generate low-affinity antibodies against any polysaccharide produced by bacteria. When adolescents are immunized with conjugate vaccines (Fig. 3C), the polysaccharide induces T cell–independent extrafollicular proliferation of the preexisting low-affinity memory B cells, which switch to plasma cells and produce a rapid immune response. The addition of adjuvants or the use of multiple doses during the primary immunization has no effect on the immune response because this depends mostly on the number of memory B cells present at the time of immunization and the ability of the conjugate to stimulate them. Eventually, the GC reaction will take place, and the resulting high-affinity memory B cells will respond strongly to a booster given 3 years later. There are no data on whether an adjuvant given with the first dose, even if not useful for the response at day 30, has an effect on boosting.

When adolescents are immunized with polysaccharide vaccines (Fig. 3D), they, too, develop T cell–independent extrafollicular proliferation of the preexisting low-affinity memory B cells, which switch to plasma cells and produce a rapid immune response. The first response is similar or slightly lower than that induced by conjugate vaccines. However, in this case, the absence of the T cell epitope precludes the selection of the B cells in the GCs, and they undergo apoptosis (15, 16). This causes hyporesponsiveness to subsequent immunizations (12). At each immunization with polysaccharide, the pool of memory B cells is depleted, and the response to a subsequent immunization with polysaccharide or conjugate is lower than that occurring in individuals who have not previously received a polysaccharide vaccine. As a consequence, concerns have been raised about the use of plain polysaccharide vaccines when there are valid alternative immunization strategies.

Critical interactions in the GC

One of the most critical interactions in the GC is the one between the BCR and the vaccine antigen. This interaction, reviewed in detail by Mesin et al. (14) for generic antigens, is adapted below to reflect the behavior of a glycoconjugate vaccine (Fig. 4).

Fig. 4 Proposed GC interactions between conjugates and polysaccharides, B and T cells.

(A) Conjugate vaccine. (B) Polysaccharide and protein are co-delivered but not covalently linked. (C) Conjugate vaccine in the presence of high concentration of soluble antibodies. (D) Conjugate vaccine in the presence of high concentration of antibodies and a high-dose vaccine.


The process starts in the light zone, where the BCR present on the surface of B cells binds the polysaccharide antigen on the membrane of FDCs and retrieves, in an affinity-dependent manner, both the polysaccharide and the covalently linked carrier protein (Fig. 4A). The B cells retrieve the antigen by applying tensile force so that they can discriminate between interactions with different affinities. The conjugate retrieved by this mechanism is processed, and the protein-derived component is presented to the T cells so that the T cells can indirectly sense the B cell affinity. B cells with higher affinity will retrieve more antigen and will receive stronger help from the TFH cells. The higher intensity of help received in the light zone will enable the selected B cells to undergo more cycles of replication in the dark zone and have a more efficient affinity maturation.

Three different scenarios may determine the interaction between the BCR and the polysaccharide and affect the immune response. The first one, shown in Fig. 4B, may explain why many attempts to use efficient delivery systems that co-deliver the polysaccharide together with the carrier protein have been a failure. Clearly, carrier proteins that are noncovalently associated or are not strongly bound to the polysaccharide, even if present in the same FDC, will not be retrieved efficiently by the BCR that binds the polysaccharide, and as a consequence, the B cells will receive a weaker T cell help or no help at all.

The second scenario is when the GC reaction occurs in the presence of free antibodies that compete with the BCR for the same antigen (Fig. 4C). In this case, the BCR may not have access to the conjugate vaccine on the surface of the FDC, and therefore, affinity maturation cannot occur. This scenario is likely to happen in adolescents, adults, and the elderly, and it may explain why a second vaccination given 1 or 2 months after the first one does not induce a response. Antibodies to the carrier protein can also interfere with the response by preventing antigen uptake by FDCs and B cells.

The third scenario is when the GC reaction occurs in the presence of soluble antibodies and a high-dose vaccine is used to overcome the presence of antibodies (Fig. 4D). The vaccine would provide enough free antigen to engage antibodies so that some of the antigens on the FDCs can interact with the BCR. This is possibly a reason why, when high doses of conjugates were used in adults, an increased immune response was observed (17).

Key lessons and conclusions

This Perspective has attempted to gather a full understanding of the behavior of conjugate vaccines in the clinic and in animal models and to link the observed behavior to the mechanistic explanation. This will allow us to better understand and predict how conjugate vaccines work and possibly learn how to fully exploit their potential. There are several questions not addressed in this paper. For instance, we do not have information on what happens at the site of injection of glycoconjugate vaccines and how they travel to the lymph nodes. In addition, we understand very little on how the primary sequence and the immune status of the carrier protein influence the response of the vaccine, and we do not understand why conjugates with outer membrane vesicle behave differently.

There are several questions that could be tested in the clinic and could lead to an improved use of conjugate vaccines (Table 1). The first one is the observation that the use of adjuvants during infant immunization may lead to earlier and improved immune response and an increased pool of memory B cells. This could lead to the use of fewer doses in infants while providing better short-term and long-term protection. Whereas non-alum adjuvants have not been widely used so far in infants, the increased experience with licensed adjuvant vaccines in adults and elderly is strengthening our confidence in these adjuvants and encourages us to use them in infants as well. Adjuvants such as MF59 have been safely used in infants for HIV vaccines and largely used in children and pregnant women during the H1N1 pandemic. An influenza vaccine adjuvanted with MF59 is also licensed for children in Canada. The development of new adjuvants composed of well-defined small molecules targeting TLR, which can be properly formulated to maximize the local immune reaction to the vaccine while minimizing the off-target systemic effects, represents a very promising solution for infant vaccines (18).

Table 1 Scientific questions on conjugated vaccines that can be addressed in clinical trials.

View this table:

The second issue is that none of the immunization schedules recommended for vaccination today elicit the maximal antibody responses to take advantage of the full potential of conjugate vaccines. Most recommended schedules induce only 10 to 35% of the maximal antibody titer that the vaccine can induce (Fig. 1). This conclusion derives from the observation that a second dose of vaccine given three or more years later induces an immune response that is 3 to 10 times higher than the primary response and, as a consequence, induces longer-lasting immunity. Therefore, boosting adolescents 3 years after the first vaccination could provide longer-lasting immunity, or using a second dose of conjugate vaccines in the elderly could enhance and prolong the protective immunity. Sometimes, it would be advantageous to achieve a higher immune response in adolescents, adults, and the elderly already exposed to a primary immunization. Unfortunately, this cannot be achieved by giving more doses or using adjuvants within the first couple of months; however, we know that this can be achieved 3 years later. The question that can be asked in the clinic is: What is the minimal interval between 2 months and 3 years necessary to allow the immune system to respond to a booster immunization? Because the lack of response to a second immunization within the first 2 months is likely to be due to the high concentration of soluble antibodies that compete with the BCR for the antigen on the surface of FDC (Fig. 4C), 5 to 8 months may be sufficient to achieve a successful booster immunization.

The third question raised by this Perspective is whether the antibody quality generated by vaccination with conjugates is different in infants where the vaccine primes the immune system and in adolescents/adults where the immune system has been primed by natural exposure. This question can be easily addressed today owing to the new technologies that allow high-throughput analysis of antibody repertoire. The observation that conjugates need to prime the immune system of infants, while they need only to boost a preexisting immunity in adolescents and adults, also suggests that to prime naïve B cells, infant vaccines need to be of very high quality and potency, while for a booster response, they only need to boost preexisting memory B cells, and therefore conjugates with lower potency may be sufficient.

Overall, the collective analysis of 40 years of experience with conjugate vaccines allowed us to find new ways of explaining some of the observed behaviors and to generate new questions that can be addressed in the clinic to improve design and use of these important vaccines.


View Abstract

Stay Connected to Science Translational Medicine

Editor's Blog

Navigate This Article