Tel.: +54-11-5238-7500, Fax: +54-11-5238-7501
Infectious Causes of Cancer
Long-lasting immunoprotective and therapeutic effects of a hyperstable E7 oligomer based vaccine in a murine human papillomavirus tumor model†
Article first published online: 24 AUG 2011
Copyright © 2011 UICC
International Journal of Cancer
Volume 130, Issue 8, pages 1813–1820, 15 April 2012
How to Cite
Cerutti, M. L., Alonso, L. G., Tatti, S. and de Prat-Gay, G. (2012), Long-lasting immunoprotective and therapeutic effects of a hyperstable E7 oligomer based vaccine in a murine human papillomavirus tumor model. Int. J. Cancer, 130: 1813–1820. doi: 10.1002/ijc.26294
This article was published online on 21 July 2011. An error was subsequently identified. This notice is included in the online and print versions to indicate that both have been corrected 16 February 2012.
- Issue published online: 22 FEB 2012
- Article first published online: 24 AUG 2011
- Accepted manuscript online: 21 JUL 2011 11:55AM EST
- Manuscript Accepted: 28 JUN 2011
- Manuscript Received: 26 APR 2011
Cervical cancer and many other anogenital and oropharyngeal carcinomas are strongly associated with high-risk human papillomavirus (HPV) persistent infections. HPV E7 oncoprotein is the major viral transforming factor, emerging as a natural candidate for immunotherapy, since it is constitutively expressed in HPV-induced cancer cells. We have previously shown that E7 can self-assemble into soluble and homogeneous spherical oligomers, named E7 soluble oligomers (E7SOs). These are highly resistant to thermal denaturation, providing an additional advantage given the demand for highly stable vaccine formulations. Here, we present a new chemically stabilized form of the E7SOs (E7SOx) and analyzed its effect in a murine HPV-tumor model. Vaccination of female mice with low doses of E7SOx combined with a CpG-rich oligonucleotide (ODN) as adjuvant elicits a strong long-lasting protection against E7-expressing tumor cells, preventing tumor outgrowth after rechallenge 90-days later. Therapeutic experiments showed that E7SOx/ODN vaccination significantly delays tumor growth and extends the time of survival of the treated mice in a dose-dependent manner. These proof-of-principle preclinical experiments denote the potential applicability of our E7SOx-based vaccine to the treatment of cervical cancer and other mucosal HPV-related neoplastic lesions. In addition to thermal, chemical and proteolysis stability, the combined recombinant and chemical modification nature of the E7SOx vaccine candidate, results in low-cost, of particular interest in developing countries, where most of the cervical cancer cases occur and the most affected population is at reproductive age.
Persistent infection by high-risk human papillomavirus (HPV) genotypes is the necessary factor for the development of cervical cancer, the second cause of female gynecologic cancer mortality worldwide. The incidence of cervical cancer on women is quadrupled in developing countries, where 80% of the cases occur. In addition to cancer of the uterine cervix, these same high-risk HPVs are also associated with other anogenital tumors and, in less proportion, to oropharyngeal cancers.1
Cervical cancer results from a spectrum of precursor lesions that are histologically very well defined, namely, the cervical intraepithelial neoplasias (CIN). These types of lesions are routinely detected in the course of cytological screening programs (Paps) and are usually removed surgically or eliminated via laser therapy or cryotherapy. Thus, drug or immunological therapies are valuable alternative or complementary options to treat both, benign warts and precancerous lesions. Two virus-like particle-based vaccines recently developed are highly efficacious for preventing the infection. However, despite their approximately 100% efficacy at preventing HPV infection, these vaccines are unlikely to have a therapeutic effect on preexisting neoplastic processes, and thus no immediate impact on the incidence of cervical cancer.2, 3 This means that effective treatment for the female world population between 15 and 75 years old is still required, where cancer deaths over the next 20 years due to persistent high risk HPV infections is estimated to be 5 millions.4
Human papillomaviruses are nonenveloped viruses containing an 8 Kb double stranded DNA genome. This genome encodes eight proteins, two of them, E6 and E7 cooperate for transformation by HPV.5, 6 The primary event of transformation by E6 is promoting the proteasomal degradation of p53.7 E7 was early described to bind the retinoblastoma tumor suppressor (pRb),8 and target it for degradation via the ubiquitin proteasome pathway.9 In any case, E7 is the major transforming protein in HPV, since its interaction with pRb disrupts growth-suppressive pRb-E2F complexes that regulate the G1/S phase transition.7, 9 Along with their oncogenic properties, E6 and E7 were also shown to participate in immune evasion mechanisms (reviewed in10). Both proteins, mostly from high-risk types, have been shown to down-regulate the expression of several IFN-inducible genes and down-regulate the expression of components of the antigen processing and presentation Pathways.
HPV16 E7 is an extended, nonglobular, dimer, conformed by two well-defined domains: an intrinsically disordered N-terminal domain,11–13 and a Zn-binding C-terminal globular domain through which dimerization occurs.11, 14, 15 Along with these properties, HPV16 E7 was shown to self-assemble into defined spherical oligomers, upon removal of its coordinated zinc atom from its C-terminal domain.16 The assembly process is very slow, undergoing substantial conformational transitions with concomitant consolidation of tertiary structure. The resulting E7 spherical oligomers (E7SOs) are highly stable, have an average molecular mass of 790 kDa and a diameter of 50 nm. They display amyloid-like properties, as they can bind thioflavin-T and congo red dyes.16 In addition, these oligomers have shown to display chaperone holdase-like activity.17
We have located endogenous oligomeric E7 species in the cytosol of three HPV16-E7 expressing cell lines, strongly colocalizing with amyloid structures, while monomeric-dimeric E7 are found in the nucleus. In all cell models tested, the cytosolic E7SOs appear as the most abundant species.18 Nuclear E7 levels do not result from protein synthesis but are replenished from the cytosolic pool. Our previous results suggest that long-term events related to de-repression of E7 would cause accumulation of E7 excess into oligomeric species in the cytosol. Consistently with this finding, E7 oligomers were detected in the cytosol of cancerous cells from tissue biopsies.18
The foreign nature of this protein together with its constitutive expression in HPV-transformed cells has positioned E7 as an ideal target for immunotherapy. In the last years, several HPV-16 E7 based vaccine candidates have been tested at preclinical and clinical stages. These vaccines cover a wide spectrum of technical approaches: DNA plasmids or recombinant live-vectors expressing the E7 gene, fusions of E7 with other proteins, peptides comprising HLA-E7 epitopes, or E7-loaded dendritic cells (DCs).19, 20
In this study, we show that E7SOs can raise long-lasting protection against tumor challenge with an established model for HPV tumors in mice. Moreover, the oligomers can revert the formation of a growing tumor, strongly suggesting its use as a therapeutic agent. Despite the fact that the E7SOs are highly thermostable, they where shown to be more resistant to proteolysis than the E7 dimer. Preclinical studies carried out with a further chemically stabilized form of the E7 oligomers, namely E7SOx, show this molecule can raise long-lasting protection against E7-expressing tumor cells and can revert the growth of implanted tumor. Overall, the results indicate that E7SOx is a strong therapeutic candidate for the treatment of HPV-associated neoplastic lesions.
Trypsin protease (proteomic grade, Sigma) was added to ninety microliters of a 45 μM solution of E7 and E7SO proteins in 20 mM Tris-HCl pH 8.0 buffer. Samples were incubated at 37°C with a 1:50 protease-to-protein weight ratio and aliquots of 15 μl (7 μg of protein) were used for each time point. The reactions were stopped by rapid cooling on ice, and by addition of the serine protease inhibitor phenylmethylsulfonyl fluoride (PMSF, 2 mM final concentration). The protease digestion products were boiled immediately for 5 min and separated in a 18% SDS-PAGE and stained by Coomassie blue.
E7SOs oxidation was carried out by incubating a 40 μM solution of E7SOs with copper sulfate up to a final concentration of 20 μM in sodium phosphate 10 mM pH 7.0.21 This mixture was incubated at 28°C for 24 hours and dialyzed against phosphate buffer to remove the copper excess. Protein concentration was determined by tyrosine absorption at 293 nm in 0.1 M NaOH (∑ tyr = 2.4 mM−1 cm−1) and UV absorption at 220 nm in HCl. The final protein solution was stored at −80°C and the degree of oxidation of E7SO sulfhydryls was analyzed by SDS-PAGE. Six micrograms of E7SOs treated with cupper sulfate (E7SOx), E7SOs and E7 proteins were incubated with or without 20 mM DTT in sample buffer (Tris-HCl 60 mM, 2% SDS, 10% glycerol, 0.01% Bromophenol Blue pH 6.8). Subsequently, all samples were heated for 3 minutes at 96°C and treated with Iodoacetamide (55 mM) to avoid reductant diffusion along running. Samples were resolved in a 15% SDS-PAGE and stained by standard Coomassie Blue. Size exclusion chromatography (SEC) experiments were performed in a Superdex S200 (GE, Healthcare Life Sciences) column in 150 mM sodium phosphate buffer (pH 7.0) plus 1 mM DTT. Five hundred microliters of 27 μM of E7SOs and E7SOx samples were injected onto the SEC column and the absorbance was recorded at 225 nm.
A cupper treated E7 oligomer sample was prepared and analyzed for electron microscopy measurement as described before.16 Briefly, a 40 μM aliquot of cupper treated E7 oligomer sample was adsorbed for 1 min onto a Formvar-coated nickel grids (Electron Microscopy Science, Fort Washington, PA), then rinsed briefly with water, and negatively stained with 4% filtered aqueous sodium phosphotungstate. Electron microscopy was performed with a Zeiss, EM 109 instrument.
Monophosphoryl Lipid A (MPL) was prepared as an aqueous formulation as described in.22 Briefly, one vial of 1 mg MPL (Sigma) was reconstituted in 980 μl of a filtered 0.5 % (v/v) Triethanolamine solution, heated at 65°C for 5 min and the pH was adjusted 7.2–7.4 with 20 μl of HCl. CpG-rich oligonucleotide (ODN) 2006 sequence: 5′-TCGTCGTTTTGTCGTTTTGTCGTT-3′ (IDT, IA). This ODN has a phosphorothioate-modified backbone to provide nuclease resistance.23 The oligonucleotide was resuspended in TE buffer (10 mM Tris-HCl pH 8.0 and 1 mM EDTA) at 3.5 mg/ml final concentration and stored at −20°C. E7SOx and adjuvant were mixed prior to each immunization.
Mice and cell lines
Six to eight-week-old female C57BL/6 mice were obtained from the animal facility of the Fundación Instituto Leloir, Buenos Aires, Argentina. All animal procedures were performed according to institutional guidelines. TC-1 cells were purchased from ATCC according to supplier instructions. The TC-1 cell line derives from primary lung epithelial cells of C57BL/6 mice, which were immortalized with HPV-16 E6 and E7 genes and transformed with the cellular c-Ha-ras oncogene.24
Tumour challenge experiments
Mice were randomly distributed into groups (n = 5) and vaccinated intraperitoneally (i.p.) twice at 21 days intervals with 50 μg of E7SOx protein and 25 μg of MPL adjuvant (E7SOx/MPL group), 50 μg of E7SOx in absence of adjuvant (E7SOx/PBS group) or 25 μg of MPL alone (MPL group). Seven days after the last boost (day 28), mice were subcutaneously (s.c.) inoculated into the left flank with 5 × 104 TC-1 cells. In all experiments, the viability of tumor cells implanted into mice was > 90%. Tumor growth was measured twice a week with a digital caliper (Schwyz), and the volume of the tumor was estimated as (length × width2)/2. Mice were killed when tumor size reached ∼3 cm3.
Groups of mice (n = 6–8) were vaccinated subcutaneously (s.c.) at the base tail with 3.75 μg of E7SOx protein and 30 μg of ODN adjuvant E7SOx vaccine (E7SOxVx group), in 100 μl final volume. A control group was injected with 30 μg of adjuvant alone (ODN group). A second boost was administered at day 14 and seven days later (day 21) mice were inoculated s.c. into the left flank with 5 × 104 TC-1 cells. Ninety days after the first challenge, all tumor-free mice were challenged again with an equal dose of TC-1 cells in the contralateral flank and monitored for another 60 days. As a control of TC-1 tumorigenicity, a group of naïve mice was included.
For therapeutic experiments, mice were first inoculated s.c. with 5 × 104 TC-1 tumor cells at the left flank (day 0). Four days after tumor implantation, mice were arbitrarily assigned into groups (8–10 per group) and vaccinated s.c. at the base of the tail with different doses of E7SOx protein (3.75 and 60 μg) combined with ODN (30 μg) (E7SOxVx-3.75 and E7SOxVx-60 groups, respectively). Control ODN group received 30 μg of ODN alone. A second boost was administered 10 days later. Mice were killed when tumor size reached ∼3 cm3.
GraphPad PRISM 4.0 software (GraphPad) was used for statistical analyses. Survival curves were estimated using a Kaplan-Meier plot and compared using the log-rank test. Statistical significance was set at p < 0.05.
Choice of the immunotherapeutic candidate species
E7SOs are not only the most abundant species in CaSki a tumor line transformed by HPV16 used as model, but also in TC-1 cells, the tumor model used to test HPV oncoprotein based vaccines in animals.18 Moreover, the oligomers appear as the most abundant species in cancerous tissue as well. In addition, there is a known correlation between the size of molecules or molecular assemblies and immunogenicity.25 Altogether, these facts made us consider the use of oligomeric E7SOs species as immunotherapeutic.
E7SOs are further stabilized by oxidation
We have already described that both the E7 dimer and E7SOs are highly thermostable proteins.11, 16 To further test the stability of the E7SOs to be used in immunization experiments in comparison with the E7 dimer, we subjected both species to limited proteolysis with the serine protease trypsin. After 5 min of mild digestion, the E7 polypeptide band (running as 22 kDa) looses intensity while lower molecular weight fragments gradually appear (Fig. 1, left panel). After 30 min incubation, E7 was readily cleaved to a metastable product of a ∼10 kDa and the 22 kDa band is lost. Unlike the E7 dimer, E7SOs show high resistance to protease degradation. The E7 monomer band persists during the entire incubation period, albeit with lower intensity. Different to what is observed with the dimer, E7SOs trypsinization does not yield a defined end product, but multiple bands of slightly lower molecular weight (Fig. 1, right panel). Moreover, the E7SOs tryptic cleavage products correspond to oligomeric species, albeit of lower molecular weight than untreated E7SOs (data not shown).
Despite the high resistance of E7SOs to proteolytic degradation, we decided to test the possibility of further stabilizing the C-terminal assembled oligomers by oxidizing their free sulfhydryl groups. It is well known that disulfide formation stabilize peptides and proteins21 and based on the existence of seven cysteine residues in the HPV16 E7 sequence we decided to use copper-induced sulfhydryl oxidation to crosslink the supramolecular assembled structure. Figure 2a shows the migration pattern of cupper treated E7SOs (E7SOx) in the presence and absence of a potent reductant such as DTT. In absence of DTT, the E7 monomer band is barely perceptible and most of the E7SOx sample remains as high molecular weight species at the top of the gel. Incubation of E7SOx with DTT reverses intersubunit disulfide bond formation, leading to an increment of the 22 kDa band. As control, untreated E7 and E7SOs proteins in the presence or absence of DTT were ran in parallel. Overall, these results indicate that the oxidation reaction of E7 oligomers with cupper sulfate is virtually complete, producing very stable, covalently linked oligomeric particles. In addition, 6.0 M concentration of the string denaturant guanidine hydrochloride does not denature E7SOx, unless DTT is added (not shown). Size exclusion chromatography and subsequent electron microscopy analysis of the eluted peak show that the E7SOx retains its morphology resembling the spherical shape observed for the E7SOs (Fig. 2b).16
Immunoprotective antitumor effect of E7SOx
To induce a potent and long-term immune response against HPV-antigen expressing tumors cells, we decided to combine the E7SOx protein antigen with the Monophosphoryl lipid A (MPL) adjuvant. MPL is a nontoxic derivative of Gram-negative bacterial cell wall lipopolysaccharide (LPS) with increasing applications in new-generation human vaccines and formulated in the recently approved prophylactic HPV vaccine Cervarix and Hepatitis B vaccine Fendrix from GlaxoSmithKline.
Figure 3 shows the percent of mice that remained tumor free after challenging with a lethal dose of the HPV-16 E7 expressing TC-1 tumor cells. As reported for the normal evolution of the tumor,24 mice receiving MPL adjuvant (25 μg/mouse) developed large aggressive tumors within 10–20 days post-implantation. In contrast, vaccination with 50 μg of E7SOx plus 25 μg of MPL resulted in the protection of 80% of mice, which remained tumor free over a 90-day follow-up. However, mice injected with 50 μg E7SOx in absence of adjuvant (E7SOx/PBS group) behaved as those receiving adjuvant alone, indicating that the E7SOx particles alone does not suffice to elicit an immunoprotective response in this HPV-tumor model.
Subsequent therapeutic experiments with the E7SOx/MPL combination were not satisfactory (data not shown). Thus, we decided to reformulate our E7SOx particles with another potent innate immune response immunostimulator: a CpG rich-oligonucleotide (ODN, reviewed in26). Similar to what has been observed before with MPL, mice receiving ODN alone (30 μg/mouse) developed tumors within 20 days, being all sacrificed by day 40 (Fig. 4). On the contrary, prophylactic vaccination with E7SOx plus ODN (E7SOx/ODN group) impaired TC-1 tumor growth in 100% of mice, being achieved with low doses of E7SOx/MPL (3.75 μg/mouse). Remarkably, all E7SOx/ODN vaccinated mice received a second tumor challenge at day 90 and remained tumor-free up to day 150. This demonstrates that low doses of the E7SOx/ODN vaccine (E7SOxVx) elicit a strong and long-lasting immunity against HPV E6/E7-expressing tumor cells.
Immunotherapeutic effect of the E7SOx vaccine
Once the protective effect of E7SOxVx against TC-1 tumor development was demonstrated, we decided to evaluate its therapeutic potential. In this case, mice were first inoculated with the TC-1 cells, and vaccinated with low and high doses of E7SOxVx four days later. A second boost was administered at day 14. Therapeutic vaccination with the same dose as that used in the challenge experiments (3.75 μg of E7SOx plus 30 μg ODN, E7SOxVx-3.75 group), significantly delayed tumor growth (Fig. 5a). Although E7SOxVx-3.75 vaccination was not sufficient to impair tumor development, it extended the time of survival by 2-fold with respect to the control group (p = 0.0017 for E7SOxVx-3.75 vs. ODN; Fig. 5b). Treatment with high doses of E7SOxVx (60 μg of E7SOx plus 30 μg ODN, E7SOxVx-60 group) completely reverted the initial tumor outgrowth in 100% of the animals, producing a significant increase in the survival time with respect to the E7SOxVx-3.75 mice (p = 0.0001; Fig. 5b). Altogether, these results show that E7SOxVx has strong antitumor effect on the E7-expressing TC-1 tumor and that this effect is dose-dependent. Nevertheless, significant therapeutic results can be obtained after treatment with low micrograms doses of the E7SOx protein.
Initial HPV infection is largely counteracted by humoral immunity in most cases, preventing viral spreading via production of neutralizing antibodies against the capsid proteins L1 and L2. However, if the initial humoral immune response fails to control or clear the infection, a persistent infection is established and a robust and specific T-cell immunity needs to be launched to avoid cancer progression, in particular in the cases of high-risk HPV type infections.
Prophylactic HPV vaccination programs with the now commercially available VLP vaccines are intended to reach the female population between 9 and 25 years old. Considering most women at risk are under reproductive age and the fact that actual therapeutic procedures entail a certain degree of associated morbidity, or can even lead to infertility, the accessibility to a therapeutic vaccine to treat established HPV infections is mandatory. HPV therapeutic vaccines would make a significant impact mainly in low-income countries, where preventive vaccination programs are still under revision, but also in developed countries where the sanitary system must cover reproductive health problems associated to previous cervical surgery procedures.
It is widely accepted that more effective vaccine approaches must stimulate both arms of the patient's immune system: the innate and adaptive immune responses. This may be achieved by the combination of a disease-specific agent with immunomodulator molecules, like pro-inflammatory cytokines or toll-like receptor ligands. CpG DNAs are strong immunostimulators, favoring T helper 1 type T-cell responses and pro-inflammatory cytokine production.26–30 This activation profile makes them suitable for therapeutic applications in chronic infectious diseases, autoimmunity and cancer. The effectivity of other HPV E7-based vaccine candidates formulated with a murine leukocyte activator CpG ODN has already been reported.31–33 However, recognition of CpG ODN motifs is known to be a highly species-specific mechanism, with strong differences in lymphocyte activation responses even within primates.23 Thus, to reduce the likelihood of loosing efficacy in translating from preclinical to clinical trials, we decided to employ here an ODN sequence containing an optimized human CpG motif with probed stimulatory effect on human leukocytes.23
We here present proof-of-principle experiments showing that self-assembled and stabilized oligomers of the E7 oncoprotein (E7SOx) with the above mentioned ODN adjuvant can be used as both, a protective and therapeutic agent against HPV induced tumors. This molecule showed 100% effectivity in preventing high risk HPV E7 transformed cells implantation and provided long-lasting immunity, impairing development of second tumor challenges. Remarkably, low doses of E7SOs also exerted a significant therapeutic effect on initial growing tumors, indicating that the E7SOxVx can be used to treat initial grade HPV-16+ CIN lesions. New experiments intended to evaluate the therapeutic effect of our vaccine candidate over larger tumors are under way as well as extending the application to other HPV types.
One reason why this particular oligomeric assembly of 790 kDa can be effective in arising a strong immune response, is precisely its increased size. A correlation of molecular size and immunogenicity of the antigen is well known,25 and the response to E7SOxVx is likely based on T-cell response, something that should be further investigated. Besides, E7SOs have chaperone activity,17 and chaperones have been known to be immunodominant antigens recognized by the immune system following microbial infection.34, 35 Further, antigen fusions with heat shock proteins can be used to induce enhanced antigen-specific response (for a review36), also in the case of an E7 vaccine candidate (under trials).37 Finally, it is worth mentioning that molecular chaperones or chaperone-peptide complexes acting as tumor specific antigens, can be used to induce tumor immunity.38 An important aspect to bear in mind is that oligomeric species of E7 are the most populated in model cell lines, including the TC-1 HPV transformed model, and more importantly, in cancerous tissue.18 Thus, E7 oligomers could be the key targets of a successful immune response, but they are hidden within cells and not circulating, thus remaining invisible to specialized antigen presenting dendritic cells.
Protein vaccines candidates must be highly stable to ensure stable vaccine formulation and long in vivo availability. At present, improvement of vaccine's stability constitutes a major challenge in the vaccine industry. Large quantities of vaccines doses must be discarded each year due to nonexistent or irregular cold chains, especially in nondeveloped countries,39 demanding the generation of vaccine components with increased temperature, pH, chemical and proteolytic resistance. Lack of stability of vaccines can lead to loss of antigenic properties, due to modification of antigen structure or conformation and disruption of antigen epitopes.39 This experiments demonstrated that E7SOxVx meet critical requisites of vaccine antigens, such as thermal stability, protease resistance and chemical stability. The latter arises from oxidation of E7SOs, which resulted in a very efficient method to link covalently E7SOs subunits by intermolecular disulfide bridges. In addition, the bacterial recombinant nature and chemical modification used for production of the E7SOx vaccine candidate provides a further important advantage related to its cost.
Early implementation of immunotherapeutic protocols to the treatment of high-risk HPV DNA positive patients may significantly improve their disease outcome. In addition, it can be a sensible complement to surgery in advanced-grade HPV-neoplastic lesions, with the further advantage of minimizing the rate of recurrence. The E7SOx vaccine approach we present here also provides the possibility to develop a combined particle with the most prevalent high-risk HPV types, targeting most people at risk. Further applications on benign mucosal and cutaneous HPV-associated lesions cannot be discarded.
The authors thank Gabriela Camporeale for technical assistance. XBio Inc. is involved in the development of vaccines. A patent application PCT/US 10/58657 for the vaccine described in the paper has been filled.
- 2Current and Future HPV Vaccines: Promise and Challenges. Seattle: PATH, 2006. 72 p., , , .
- 5Papillomaviruses and their replication. In: Knipe DM, Howley PM, eds. Fields Virology, 4th edn. Philadelphia: Lippincott Williams & Wilkins, 2001. 2197–229., .
- 21Role of disulfide bonds in folding of recombinant human granulocyte colony stimulating factor produced in Escherichia coli. In: Cleland JL, ed. Protein Folding: In Vivo and In Vitro, vol. 526. 1993. 189–202., , , , .
- 36Heat shock protein fusions: a platform for the induction of antigen-specific immunity. In: Henderson B, Pockley AG, eds. Molecular chaperones and cell signalling. New York: Cambridge University Press, 2005. 288–99., .
- 38Molecular chaperones as inducers of tumour immunity. In: Henderson B, Pockley AG, eds. Molecular chaperones and cell signalling. New York: Cambridge University Press, 2005. 300–17., .
- 39Vaccine manufacturing. In: Plotkin SA, Orenstein WA, eds. Vaccines, 4th edn. Philadelphia: Elsevier Inc., 2004. 53–67., .