review of literature hepatitis b is the most serious liver...
TRANSCRIPT
Review of Literature
20
2.1 Hepatitis B disease
Hepatit is B is the most serious liver infection in the world and can
lead to liver failure, c irrhosis or cancer of the liver (Hepati tis B
Foundation). It is often characterised by jaundice, abdominal pain, l iver
enlargement and fever. According to the WHO, it is estimated that 400
mil l ion people worldwide are already chronically infec ted with hepatit is
B and th is disease leads to over 1 mill ion deaths per year. This disease
is transmitted by blood and body f luids, frequently during sexual
intercourse. Treatment is generally ineffective (Hepatit is B Foundation)
and thus prevention and vaccination remain the best a lternatives for
this disease.
Hepatit is B is caused by the hepatit is B v irus (HBV), an envelope
virus contain ing a partially double s tranded, circular DNA genome and
classif ied within the family hepadnav irus (Shepard et al. , 2006). The
virus interferes with the function of the liver in replicating hepatocytes
caus ing the immune sys tem to be activated leading to liver
inflammation.
The Hepati tis B v irus is a spherical partic le with a diameter
of 42 – 47 nm and circulates in the blood in concentrations as high
as 108 v irions per mL (Shepard et al., 2006). The partic les are
composed of an envelope, which consis ts of proteins known as HBs or
surface prote ins. These self-assembling units form non-infectious
spherical or tubular partic les (Robinson and Lutwick, 1976) containing
Review of Literature
21
the surface antigens (HBsAg) which are respons ible for elic it ing the
immune response. These HBsAg partic les contain only envelope
glycoproteins and host-derived lipids and outnumber v ir ions by 1000:1
to 10,000:1 (Ganem and Prince, 2004). The surface coat surrounds an
inner protein shell , composed of HBc proteins. Final ly, the v iral DNA
and enzyme DNA polymerase can be found in the inner most region of
the partic le.
2.2 Structural Components of HBV
Infection by HBV results in secretion of high t iters of an antigen
called Australia antigen in the sera (Dane et a l., 1970) on three types
of partic les: (1) pleomorphic spheres of approximately 20 nm diameter,
(2) f i laments of variable length and diameter of 20 nm, and (3)
spherical double shelled partic les of 42 nm that are referred to as Dane
partic les . The inner protein shell of the Dane partic le is called the core
partic le, which contains core antigen (HBcAg). The antigen present on
the outer shell of the Dane partic le is referred to as hepatit is B surface
antigen (HBsAg). The core antigen surrounds the v iral genome, a small
DNA molecule of unusual s truc ture. Another antigen, which is a
derivative of the core antigen and known as e antigen (HBeAg) is found
in various forms (Mackay et a l., 1981). Among the three antigens,
HBsAg was found to be protective antigen suitable for vaccine
development.
Review of Literature
22
2.3 HBV Structure and Pathogenicity
The Human HBV virus consists of a part ially double-stranded DNA
genome of 3.2 kb enclosed by envelope proteins (HBsAg). The genome is
packaged wi th a core protein (HBcAg) and a DNA polymerase. Fol lowing the
receptor-mediated entry into a hepatocyte, it enters in to the nucleus and i ts
DNA becomes integrated into the host genome. Prote in synthesis proceeds
f rom four open reading f rames (ORFs): the envelope proteins (large, middle
and major HBsAg) f rom the S gene, pre-S1, and pre-S2 gene sequences; the
e antigen (HBeAg) and HBcAg f rom the C gene and pre-C gene sequence;
the DNA polymerase pro tein f rom the P gene; and the transactivator X
protein f rom the X gene (Fig. 2.1). DNA replication proceeds v ia RNA
intermedia tes in the nucleus. The vi rus part icles are then assembled in the
cytoplasm and re leased by the hepatocyte (Pugh and Bassendine, 1990).
Al though the integration of the vi ral genome is not necessary for i ts
repli cation, i t allows the persistence of the v i ral genome in the cells of host
(Tiollais et a l ., 1985; Ganem and Varmus, 1987).
Fig. 2.1: Hepatitis B virus (HBV) genome organization
HBV is a member of the 'hepadnavirus' group, double-stranded DNA viruses, which repl ica te, unusual ly, by reverse transcription of an RNA in termediate, the pregenomic RNA. The f igure shows the physical map of the HBV
Review of Literature
23
genome. The genome is approximately 3200 nucleotides in length . HBV vi rions contain both DNA and RNA and some regions of the packaged genome can be single stranded, double stranded or even triple stranded. There are four def ined overlapp ing open reading f rames (ORFs) in the genome which resul t in the transcript ion and expression of the seven dif fe rent hepatit is B proteins through the use of varying in-f rame start codons. The four ORFs transcription are controlled by four promoter elements (preS1, preS2, core and X), and two enhancer elements (Enh I and Enh II). Source: J Viral Hepat 2004.
The core and po lymerase genes are essential for v i ral DNA replica tion,
and the envelope proteins are essential for envelopment of nucleocapsids.
HBx pro tein and HbeAg are expressed during natural infect ions and their
funct ion is not too c lear. HBx pro tein has been seen to be required for the
establ ishment of an infection in vivo (Chen et al . , 1993; Zoulim et a l ., 1994)
but is dispensable for v i rus replica tion in transfected ce lls (Blum et al . , 1992;
Yaginuma et al . , 1987).
2.4 The HBV replication cycle
Despite 40 years of HBV research no widely available cell l ines
permissive for HBV or any other member of the Hepadnav ir idae family
has been described. Studies on the repl ication cycle of
Hepadnavir idae , i.e., attachment, entry, genome replication,
transcription and expression of v iral genes, assembly, and budding
cannot be fully studied or are limited to small series of experiments
with primary permissive hepatocytes (Aldrich et a l., 1989; Gripon et al. ,
1993; Gripon et al., 1988; Ochiya et al., 1989; Tutt leman et a l., 1986).
Unfortunately, these primary hepatocytes remain permiss ive for only a
short t ime after being obtained from the intact l iver.
Review of Literature
24
It is assumed that v irus entry and the host range of
Hepadnav ir idae is dependent on the N terminus of the large surface
antigen (Chouteau et al., 2001; Gripon et al., 2005; Ishikawa and
Ganem 1995; Lambert et a l., 1990; Verschoor et al., 2001). So far, the
intr insic HBV receptor has not been discovered, but from studies on
DHBV in primary duck hepatocytes it is assumed that around 104
receptor molecules per cell mediate the rapid binding, followed by a
slow uptake of the v irus to the cel l which can take up to 16 hours
(Hagelstein et a l., 1997; Kl ingmuller and Schaller, 1993; Kock et a l.,
1996; Pugh et al., 1995; Pugh and Summers, 1989; Rigg and Schaller,
1992). Following entry into the hepatocyte and uncoating, which may
proceed in parallel , the nuc leocapsid is transported into the cell’s
nucleus where the v iral nucleic acid is released. Release of the v iral
DNA and disintegration of the nuc leocaps id is assumed to take place at
the nuclear pore complex (Kann et a l., 1997; Rabe et a l., 2003).
In the infected hepatocyte the viral DNA is immediately
transformed into the covalently closed circular (ccc) DNA by cellular
enzymes. The cccDNA in turn is the template for transcription of v iral
genes, and acts chemically and structurally l ike an
episomal/extrachromosomal DNA and has a plasmid-like structure
(Bock et a l., 1994; Bock et a l., 2001; Newbold et a l., 1995). Congruent
with the fact that HBV infec ts hepatocytes, nearly all e lements
regulating v iral transcription have binding sties for l iver-specif ic
transcription factors (Courtois et al., 1987; Guo et al., 1993; Lopez-
Review of Literature
25
Cabrera et a l., 1990; Lopez-Cabrera et a l., 1991; Raney et a l., 1995;
Schaller and Fischer, 1991). Nevertheless, although a number of
factors and interactions regulating v iral transcription are known, the
exact mechanisms of HBV transcription remains unclear. However, v iral
transcription occurs in the nucleus, and both messenger and
pregenomic RNAs are transported into the cytoplasm where they are
translated or used as the template for the production of progeny
genome, respectively.
In the cytoplasm, the core protein which i tself can be
phosphorylated by several kinases forms the basis for the nucleocapid
and plays an ac tive role in binding and packaging of the pregenomic
RNA, recruitment of the v iral polymerase, and thus enables the RT
polymerase/RNA complex to init iate reverse transcription within the
newly forming nuc leocaps ids (Daub et al., 2002; Gerlich et a l., 1982;
Kann et al., 1993; Kau and Ting, 1998; Lan et al., 1999; Liao and Ou,
1995; Watts et al., 2002).
The three surface proteins of HBV have two major properties:
First, as transmembrane proteins they are anchored in the v iral
envelope and thus are located on the surface of the virus, being
respons ible for binding to the so-far-unknown v iral receptor. Second,
the three surface proteins are secreted as subviral partic les that do not
contain a functional nucleocapsid. The proteins di ffer in their N-
terminal sequences that are longer in case of the L and M protein. All
Review of Literature
26
proteins have in common the S domain, M addit ional ly has the preS2
domain, L has both the preS2 and in addition the preS1 domain (Figure
2). The surface proteins of mammalian Hepdnav ir idae have been shown
to be N- and O-glycosylated. These glycosylations have been shown to
be responsible for proper secretion of progeny v iral partic les and in
turn may represent novel targets for therapies with glycosylation
inhibi tors (Block et al., 1998; Block et al., 1994; Lu et al., 2003;
Schildgen et al., 2004; Schmitt et al., 1999; Schmitt et a l., 2004).
Moreover, the surface proteins have been demonstrated to be
activators of transcription by acting in trans (Caselman et a l., 1990;
Kekule et a l., 1990).
The v iral polymerase, s ingle most enzyme encoded by the
hepadnav iral genome, consists of three functional domains – the
terminal protein, the reverse transcriptase, and the RNaseH domain –
and a spacer domain that separates the terminal protein domain from
the polymerase domains. The terminal protein also serves as a primer
for the reverse transcription (Lanford et al., 1997; Wang and Seeger,
1992; Weber et al., 1994). Before or during formation of the cccDNA
the terminal prote in but also one of the redundant terminal repeats
present on the relaxed circular v iral genomic DNA that is released from
the nucleocapsid are removed and the cccDNA forms by not fully
understood mechanism, most probably dependent on cellular l igases
and maybe further enzymes. So far i t is assumed that cellular DNA
Review of Literature
27
repair mechanisms become active and convey the relaxed circular form
into the cccDNA (Seeger et al., 2007).
The cccDNA is the template also for the pre-genomic RNA
(pgRNA). This RNA is both the template for core and polymerase
protein trans lation but also the matr ix for the progeny genomes. The
pgRNA bears a secondary s truc ture - named e-structure - that is
present at both the 5’- and the 3’-ends. The e-hairpin loops at the 5’
end are f irst recognized by the v iral polymerase and act as the init ial
packaging signal (Bartenschlager and Schaller, 1992; Hirsch et al.,
1990; Huang and Summers , 1991). The synthesis of the DNA minus
strand, i .e. the intr ins ic reverse transcription, is then init iated by the
formation of a covalent bond between the tyrosine Y65 res idue of the
terminal protein domain and a desoxy-guanosine-monophosphate
(dGMP) (Lanford et a l., 1999; Wang and Seeger, 1992; Weber et al.,
1994; Zoulim and Seeger, 1994). The next few nucleotides following
this init ia l dGMP are complement to a small part of the e-structure. The
small terminal protein bound primer is subsequently translocated to the
3’ end v ia an unknown mechanism but remains covalently bound during
the whole t ime. Perhaps this process is a prerequisite for the correct
folding of the progeny genome within the newly forming nucleocapsid.
Finally, the minus strand is fully synthesized by the reverse
transcription reaction while the RNA is degraded by the RNaseH
activ ity of the enzyme. The fo llowing plus s trand synthesis is init iated
by an 18mer capped RNA ol igo that remains from the 5’ end of the
Review of Literature
28
pgRNA (Lien et a l., 1986; Loeb et a l., 1991). Nevertheless, it is
assumed that, whi le not actively replicating and even with conflict ing
data on its stability, there is ev idence that cccDNA may be stable in
infected hepatocytes, thus contr ibuting to chronic HBV infection,
leading to a need for long-term therapies to help eliminate the cccDNA
posit ive cells.
The final replication s tep, i .e., assembly and release of Dane
partic les , is not fully understood, although from one study on usage of
glycosylation inhib itors that at nontox ic doses suppress viremia in
WHV-infected woodchucks there is indirect evidence that assembly and
release occur v ia secretory pathways (Block et a l., 1998).
2.5 Hepatitis B Surface Antigen
The HBV genome, HBsAg encodes the S region, further consists of
two regions, preS (pre S1 and pre S2) and S, with an ini tiation codon in
each region. The three forms of HBsAg are, smal l (SHBs), middle
(MHBs), and large (LHBs). Each of the three proteins encoded by
preS/S ari ses from separate trans lation init iation at each of the three in-
frame s tart codons. The S region which encodes the smallest SHBs
exis ts in glycosylated (gp 27) and unglycosylated (p24) forms. The pre
S2/S domain encodes the middle MHBs consists of a hydrophilic N-
terminal extens ion of 55 aminoacids aris ing from translation of the pre
S2 region. It exis ts as mono (gp33) and diglycosylated (gp36) forms.
Production of LHBs is resulted with an addit ional 108 or 109
Review of Literature
29
aminoacids by the translation at the from the pre S1 region. LHBs
exis ts as glycosylated (gp42) and unglycosylated (p39) forms (Gerlich
and Bruss, 1992). SHBs are the major component in all three partic les
(Fig. 2.2).
Fig. 2.2: Structural organization of the HBV Antigen ORF-S that encodes the three forms of surface antigen proteins Source: International Journal of Medical Sciences 2005; 2:2-7
SHBs carry at least one neutra lizing epitope. Two other
determinants have been described for HBsAg (Gerlich and Bruss,
1992). One determinant has either d or y specif ic ity and the other has
either w or r specif ic ity. All combinat ions of determinants are found,
result ing in four subtypes of HBV: adw, adr, ayw, and ayr (Peterson et
al., 1984). HBV strains are further class if ied in to nine serotypes, ayw
1, 2,3,4, ayr, adw 2,4, adrq-and adrq+, on the basis of their antigenic
determinants and sub-determinants of their HBsAg (Courouce et a l.,
1976; Courouce-Pauty et a l., 1978). The occurrence of escape mutants
in many regions of the world with di fferent SHBs subtypes suggests
that subtypespecif ic immunity may not play an important role if SHBs is
the only vacc ine component. Intramuscular injec tion of serumor yeast-
derived recombinant HBsAg in healthy indiv iduals results in effective
immunizat ion and protection from v iral infection (Emini et a l., 1986).
The technical adv isory group of the World Health Organization (WHO)
Review of Literature
30
has recommended the addit ion of hepatit is B vaccine in the Expanded
Programme of Immunizat ion (EPI) in al l countr ies with moderate to high
endemicity of infection (WHO, 1988; Calandra et al., 1992).
2.6 Serological heterogeneity of hepatitis B surface antigen
The serological heterogeneity of hepatit is B surface antigen
(HBsAg) has long been established. All known serotypes of HBV
contain the common a determinant and one of each of the mutual ly
exc lusive determinants d/y and w/r (Le Bouv ier, 1971; Bancroft et al .,
1972). Additional serological specif ic it ies, originally designated as
subdeterminants of a and subsequently as subdeterminants of w, have
allowed the identif ication of four serotypes of ayw and two of adw
(Courouce et al., 1976). Thus, eight subtypes of HBsAg, aywl, ayw2,
ayw3, ayw4, ayr, adw2, adw4 and adr, have been serologically defined,
also designated as P1 to P8 (Courouce et al., 1976). The q determinant
was original ly found to be expressed on HBsAg of al l subtypes apart
from adw4 (Magnius et al., 1975). Subsequently, lack of q was also
demonstrated in some adr subtype-containing sera. Thus adr strains
can be defined as either adrq+ or adrq- (Courouce-Pauty et a l., 1978).
2.7 Pathogenesis of hepadnavirus infections
The transmiss ion of HBV and other members of the
Hepadnavir idae fami ly occur both vertical ly and horizontally v ia body
fluids . A maximum of 1010 to 1012 genome copies per ml serum or
body f luid can be found. In chronic infections, the v iremia is subjec t to
Review of Literature
31
natural f luc tuations of one log (Schi ldgen et al., 2006). The rate for
chronic ity, depending on the study, is >90% in neonates and
approximately 10-15% in adults. The r isk for trans fusion-acquired and
nosocomial infections in the past two decades decreased due to
optimized molecular diagnostics and more s tr ict hygiene and legal
regulations , although there is stil l a remarkable number of such
transmission caused by incautious behav ior of healthcare personal.
Once having entered the host, the Dane partic le reaches its major
target cell, the hepatocyte, the main site for repl ication and
persistence, as vir tually all hepadnav iruses display a pronounced and
distinct l iver tropism. Furthermore, some other cell types have been
shown to serve as non-hepatic reservoirs for mammalian
hepadnav iruses. Within the infected liver in immunocompetent hosts
there is a continued damage of infected hepatocytes by cytotoxic T
lymphocytes (CTLs) that leads to uninterrupted expression of collagen
fibres that in the wors t and untreated cases lead to liver cirrhosis (Liaw
et al., 2004; Mathew et a l., 1996; Maynard et al. , 2005;
Papatheodoridis et a l., 2005; Pinzani, 1995; Rizzetto et a l., 2005;
Rockey, 2005; Yoshida et a l., 2004).
It is worth note that there is s til l no evidence that HBV is cytotoxic
for the infec ted hepatocyte. In contrast to other v iruses that can infect
the liver l ike the herpes s implex v irus (HSV), HBV is unable to induce
cytopathic effects under normal infection condit ions (J ilbert et al.,
Review of Literature
32
1992; Kajino et al., 1994; Thimme et al., 2003; Wieland et al., 2004).
Liver damage (f ibrosis, c irrhosis, and probably hepatocellular
carcinoma) is believed to be induced by the ongoing immune reaction
and the steady s tate inflammation in the liver. Consequently, confirmed
by experimental data (Ando et a l., 1994; Guidotti et a l., 1994; Guidot ti
et al., 1999; Guidotti et a l., 1994b; Guidotti et a l., 1996; Guidott i et a l.,
2000; Guidotti et al., 1999b; Kakimi et al., 2001; Tsui et al., 1995), it is
generally assumed that massive CTL and NK T cell action result ing in
the kil l ing of infected hepatocytes is essential for elimination of the
infection. It is further assumed that in those cases in which chronic
infection evolves, the init ial cellular immune response is too weak and
thus not suffic ient to control the infec tion (Ganem et a l., 2004). It
remains unclear what mechanisms are responsible for the passage
from acute to chronic infection, thus this part of the v iral l ife cycle
remains a matter of speculation. As a matter of fac t it has been shown
that a suff ic ient Th1 response involving CD8 posit ive CTLs, natural
ki l ler T cells (NK T), cytokines (TNF-alpha, interferon gamma), and
cytokines like IL-12 and IL-15 and many others are involved in the
suff ic ient suppression of trans ient infections (Seeger et al., 2007).
Despite the fact that only antibodies against the s protein are
neutralizing and are the only markers of immunity it was hypothes ized
that transient infection is kept in check by gamma interferon and other
cytokines re leased by immune cells, leading in turn to a shutdown of
v iral replication (Pasquetto et al., 2002; Schulz et a l., 1999; Schulz et
Review of Literature
33
al., 1999b). However, this does not expla in why only in those patients
who c lear the v irus HBsAg antibodies are present. This is assumed to
be a continuous control of the infection as cccDNA can be found in
these patients decades later (Maynard et al., 2005; Werle-Lapostolle et
al., 2004), while this is not the case if the infection passes to a chronic
stage.
2.8 Brief history of treatment for the Hepatitis B disease
Hepatit is B vaccines have been available since the early 1980s.
The first vaccine (HEPTAVAX®), l icensed in 1981, was a subunit
vaccine isolated from the blood of hepatit is B infected donors (Prince,
1982). In hepatit is B patients suffer ing from acute infec tion and in
hepatit is B carr iers, 22nm partic les containing HBV surface antigen
subunits of the v irus are found expressed in copious amounts. These
partic les themselves are not infectious but are effective in s timulating a
strong immune response (Dekleva, 1999). Since these partic les are
derived from blood, extensive purif ication is cr it ical to ensure that the
f inal product exceeds 99% purity and is free from extraneous infectious
viruses.
There were many l imitations in the manufacture of plasma-derived
hepatit is B vaccines. Primarily, a constant l imitation exists in the
supply of human plasma. Due to the intensity in the manufacturing
process, to produce just one batch, more than 1 year is required
(Dekleva, 1999). During the 1980s there was also growing concerns
Review of Literature
34
about AIDS infection through blood derived products (Francis et al.,
1986) that led to give a pressure to develop a recombinant vaccine for
human use.
RECOMBIVAX HB® from Merck was the f irs t recombinant hepatit is
B VLP vaccine. Recombinant hepatit is B vacc ine, ENGERIX B® w as
produced by GlaxoSmithKline (GSK), which is also currently in the
market. In addit ion, there is now a vaccine for both hepatit is A
and B together, known as TWINRIX®, which is bivalent in
nature, containing antigenic components used in producing HAVRIX®
(Hepatit is A inactivated vaccine) and ENGERIX-B® (Hepatit is B,
recombinant vaccine), by GlaxoSmithKline (GSK).
2.9 Hepatitis B Surface Antigen (HBsAg) VLP vaccine
The firs t commercially manufac tured recombinant hepatit is B VLP
licensed in 1984 (Hil leman, 2001; Hil leman, 2003; McAleer et a l. , 1984;
Valenzuela et a l. , 1982) was cloned and expressed in S. cerev isiae
(McAleer et a l., 1984; Miyanohara et al. , 1983). The VLP is assembled
from up to 100 copies of the hepatit is B surface antigen monomer and
host-derived lipids forming highly immunogenic spherical partic les
measuring ~ 22nm in diameter (Fig.2.3a). The VLP is widely referred to
as the hepatit is B surface antigen (HBsAg) after its or igins . The
comparison of the HBsAg produced by recombinant yeast for HBsAg of
human blood plasma has been reported (Yamaguchi et a l. , 1998).
Review of Literature
35
In S. cerevisiae, by the insertion of the hydrophobic segments of
the HBsAg monomer into the endoplasmic reticulum (ER) is init iated by
the assembly of the recombinant HBsAg partic les (Zhou et al. , 2006)
which then bud into the lumen of the ER as 20-nm l ipoprotein partic les
using host-derv ied machinery and l ip ids (Eble et a l. , 1986). In humans,
the HBsAg maturation process will proceed through the Golgi and the
HBsAg partic les are subsequently secreted. In yeast cells however, due
to the lack of protein transport machinery (Biemans et al. , 1992), the
HBsAg partic les are permanently localised in the ER.
(a) (b)
Fig. 2.3: (a) Cross-section of an HBsAg particle (b) Interaction between a p24s protein monomer and the lipid bilayer on the
surface of the HBsAg particle Source: (a)Koist inen (1980); (b)Mahoney and Kane, 1999.
The recombinant HBsAg partic les are cys teine-rich and the
structural evolvement and maturation depends on the level of
disulphide formation. The formation of correct intra- and intermolecular
disulphide bonds are cr it ical as they determine the conformational
stabil ity and immunogenicity of HBsAg vacc ines (Guerrero et al. , 1988;
Review of Literature
36
Zhou et al. , 2006). HBsAg derived from human blood plasma are
already in the mature form but in recombinantly expressed HBsAg, the
disulphide bond formation and the cross-l inking process continues
during purif ication. Addit ional s teps of treatment with potassium
thiocyanate (KSCN) and storage at elevated temperatur es have been
included at the end of the purif ication chain to complete the maturation
process of recombinant HBsAg (Zhou et al. , 2006).
HBsAg characterisation studies have revealed that the partic le is
composed of ~ 75% protein and ~ 25% lipid by mass (Dreesman et a l. ,
1972; Gav ilanes et al., 1982). The lipid-protein interac tions are
respons ible for the formation of the desirable helical structures of the
proteins forming the epitopes (Gav ilanes et al. , 1990) (Fig. 2.3b).
Sterol accounts for 30% of the lipid constituent of the HBsAg partic le
while phospholipids account for the remaining 70%, of which
phosphatidylcholine (PC) is the most abundant. Negatively charged
phospholipids, phosphatidylserine (PS) and phosphatidylinositol (PI), in
particular, have been shown to have s ignif icant influence on HBsAg
activ ity (Gomez-Gutierrez et a l. , 1994). A ful l account of the role of
l ipids in maintaining the structure and antigenicity of HBsAg and the
effects of l ipid reconstitution is available elsewhere (Gav ilanes et al. ,
1990; Gomez-Gutierrez et a l., 1994).
The intr ins ic charac teristics of the HBsAg partic les , outlined
above, are cri tical considerations in ensuring product quality and
Review of Literature
37
potency. Like other VLPs, HBsAg partic les are large macromolecular
entit ies which are diff icult to characterise and are susceptible to
conformational and biological changes in d ifferent microenv ironments.
As prev iously d iscussed, the process of producing the vaccine itself
often defines the end product (Buckland, 2005) and regulatory demands
are str ingent considering that the vaccine products are adminis tered to
healthy indiv iduals. It is required that assurance of the identity and
composit ion of the adminis tered dosage is provided to meet the
demands of product safety and eff icacy. Hence, in this study, the
biological and phys iochemical aspects of the HBsAg sys tem are
carefully considered alongside the purif ication process development
efforts. Ul tra-scale down (USD) techniques are particular ly valuable
here in simulating industria l scale condit ions to allow the
characterisation of the biochemical and biophysical responses of the
HBsAg partic les to different processing env ironments .
2.10 Hepatitis B Surface Antigen (HBsAg) platform for novel
vaccines
In addit ion of their use as direct immunogens, VLPs have shown
enormous potential as platforms for the development of future
generation vaccines based on hybrid or chimeric VLP. VLPs are highly
eff ic ient in stimulating cellular and humoral responses as they prov ide
the spatial structure for the display of conformational epitopes that
mimic the native v iral structure leading to enhanced antibody
production (Grgacic and Anderson, 2006).
Review of Literature
38
The hepatit is B surface antigen (HBsAg), with the abil ity to sel f-
assemble with host-derived l ipids into empty non-infec tious VLPs
(Cheong et al. , 2009), is attractive as a delivery platform for foreign
epitopes. The eff ic iency of the HBsAg partic les in pr iming cellu lar and
humoral responses even in the absence of an adjuvant has been
reported (Boisgerault et al. , 2002). Chimeric vaccines for Hepatit is C
(Netter et al. , 2001), dengue (Bisht et al. , 2001) and HIV (Michel et al. ,
2007; Greco et al. , 2007) have been developed using this platform.
The HBsAg partic les are composed of a lipid b ilayer surrounded by
external hydrophobic loops which could be genetically engineered to
carry the heterologous antigens for delivery (Delpeyroux et al. , 1986;
Netter et al. , 2001; Phogat et a l., 2008). The ability of the hydrophobic
loops to accept peptides facilitates the insertion of selec tive epitopes
agains t diseases including HIV-1 and hepatit is C (Lee et al. , 1996;
Schlienger et al. , 1992; Boisgerault et al. , 2002).
The lipid-envelope structure of the HBsAg is an attractive attr ibute
as it allows the incorporation of foreign antigens requiring the support
from a lipid bilayer structure. This aspect is particular ly relevant in the
development of HIV vaccines where a lipid membrane env ironment is
necessary to orientate the antigenic epitopes for eff ic ient induction of
broadly neutral is ing antibodies (Grunder et a l., 2002). Preliminary
studies showed that the HIV epitopes could be appended to the C-
terminus of the HBsAg S-1 protein and the l ipid env ironment prov ided
the support necessary for effec tive antibody binding (Phogat et a l.,
Review of Literature
39
2008). Simi larly, the chimeric VLP for dengue v irus envelope protein
showed enhanced immunogenic ity when assembled on the HBsAg
platform (Bisht et al. , 2002).
HBsAg partic les have a hollow core with an encapsulation
space of 900-8000 nm3 (Reimann et al. , 2006) and access to the
inter ior is mediated by pores in the bilayer. The hollow structure of the
HBsAg partic les could also be employed for the encapsulation or
entrapment of antigenic proteins and peptides , oligonucleotides or
cytokines which would be exposed to the surface by v ir tue of the pores.
Early proof-of-concept s tudies for this HBsAg appl ication have been
demonstrated (Reimann et a l. , 2006).
2.11 Production and purification methods for HBsAg
2.11.1 HBsAg expression
The HBsAg partic les used as the VLP model in this study was
developed by Merck (West Point, PA, USA) us ing S. cerevisiae as the
host cell expression system. Detai ls of the plasmid and strain
construc tion are reported elsewhere (Carty et a l. , 1989; Hinnen et a l. ,
1978).
The crucial start ing point in the development of a VLP vaccine is
the design of a cell based system which is s table over many
generations and produces high t itres of the VLP product (Buckland,
2005). The choice of the expression system is particular ly important for
l ipid-envelope VLPs for the reason that the lip id constituents are
Review of Literature
40
derived from the host cel l (Betenbaugh et al. , 1995; Buonaguro et al.,
2005). Although the production of HBsAg has also been achieved using
bacteria (Shu et al., 2006), transgenic plants (Kumar et al. , 2005) and
mammalian cel ls (Diminsky et a l. , 1997), S. cerev isiae remains a highly
attractive host. Large scale fermentation is wel l established, highly
reproducible and amenable by Yeast fermentation (Buckland, 2005).
Addit ional ly, yeast cells have the abili ty of performing complex
eukaryotic- like post-translational modif ications producing proteins
similar to those of the mammalian origin and a wealth of genetic
information is avai lable for this strain wh ich facilitates the engineering
of the desired cell metabolism and expression characteristics for the
VLP.
Fermentat ion condit ions for VLP production would depend on the
host express ion system chosen. The recombinant HBsAg used as the
VLP model in th is study is transcribed under the control of the GAL-10
promoter in S. cerevisiae. Hence, HBsAg expression is regulated by the
levels of glucose and galac tose in the cul ture media. Fermentation
conditions and glucose-galactose interactions in HBsAg expression
have been reported prev iously (Carty et a l. , 1987; Carty et al. , 1989).
For plasmid selection and maintenance, the Merck HBsAg strain also
carr ies a Leu+ gene in the same plasmid as the HBsAg gene. This
allows the use of a leucine-free basal media for fermentation which
would select for cells containing the plasmid with the Leu+ and HBsAg
genes. Fermentation for HBsAg production is typically carr ied out in a
Review of Literature
41
batch process (Carty et al. , 1987) but the use of continuous
fermentation has also been reported (Fu et al. , 1995).
An issue for cons ideration when developing yeast fermentat ion
media is that yeast cel ls are susceptible to catabolite repression, a
condition where cel l growth is hindered when excess sugars such as
glucose are present (Gancedo, 1998). Oura (1974) developed a fully
defined media for yeast fermentation us ing glucose that has minimal
catabol ite repression effects. Yau (2005), in her MEng research project
in Biochemical Engineering (UCL), proposed a media for HBsAg
production which has the combined advantages of the Oura (1974) and
the Fu et a l. (1995) media. This media was shown to be successful for
the fermentation of wild type S. cerev isiae cells but was not tes ted for
the recombinant HBsAg s train due to t ime constraints. Joyce et al.
(1998) reported on a recombinant S. cerev isiae fermentation media
which is semi-defined for the production of the human pappilomav irus
(HPV) vaccine. It is known that the S. cerev isiae host expression
system for this vaccine is similar to that for hepatit is B and hence the
media may be employed for HBsAg production.
In this research, the recombinant S. cerevisiae for HBsAg was
donated by Merck (West Point, PA, USA) in the form of a master seed
stock. Fermentation is not a key element in th is projec t however it is
crucial to produce suff ic ient feed material for subsequent purif ication
studies which were representative of material in an industr ial process.
Review of Literature
42
There is l imited information in the publ ic domain on fermentat ion media
for HBsAg production us ing the Merck recombinant S. cerev isiae strain.
To minimise t ime and research effort in generating feed stock for
purif ication studies, prior knowledge from the Yau (2005) and Joyce et
al. (1998) studies were used as the basis for fermentation
development.
2.11.2 Recovery and purification of HBsAg
For recovery of the intracellular HBsAg product, the yeast cells
following fermentat ion are harvested, concentrated and washed with
buffer for the removal of media components and anti foam. This is
usual ly accomplished us ing centr ifugation or microfiltration and the
cells are stored as cell pas te at -70 oC (Sitr in and Kubek, 1992).
Product recovery is performed by suspending a batch of frozen
cells in a buffer containing phenylmethylsulfonyl f luoride which is a
protease inhibitor (Dekleva, 1999), and the suspension is subjected to
cell disruption by homogenisation. Following cell breakage, the crude
extract is treated with a phosphate buffer containing a detergent,
usual ly Triton X-100 (Wampler et a l. , 1985). This is a c rit ical step in the
recovery of HBsAg as the detergent facil itates the liberation of the
HBsAg from tightly associated endoplasmic reticulum (ER) membrane
components (Dekleva, 1999). Removal of cell debris can be performed
by centri fugation (Wampler et a l. , 1985) or microfiltration (Deklava,
1999). Res idual detergent is removed by recirculation through
Review of Literature
43
polystyrene XAD-4 beads and the product stream can be ultrafiltered
through a 100-kDa membrane to clear small molecular weight
contaminants and for product concentration.
Purif ication is typically accomplished by chromatography-based
operations (Dekleva, 1999; Wampler et a l. , 1985; Belew et a l. , 1999)
although the use of prec ipitation and density ultracentr ifugation has
been reported (Deml et al. , 1999). The HBsAg product is init ially
purif ied v ia adsorption and elution from colloidal s il ica (Aerosil)
(Wampler et a l. , 1985). Condit ions for this process are reported
elsewhere (Pillot et al. , 1976; Sitr in and Kubek, 1992). Final pol ishing
is performed using hydrophobic interaction chromatography using butyl
agarose (Dekleva, 1999).
As highlighted by Zhou et a l. (2006), the HBsAg produced in S.
cerevisiae is not in the fully d isulphide-bonded form as found in blood-
derived HBsAg. The maturation process for recombinant HBsAg
progresses through the purif ication process. To facili ta te the format ion
of disulphide cross-linkages in the mature form, the HBsAg partic les
are treated with th iocyanate and incubated at an elevated temperature
of 37 oC prior to product formulat ion (Zhou et a l. , 2006). The product is
f inally adjuvanted by co-prec ipitation with a luminium hydroxide
(Dekleva, 1999).
Review of Literature
44
In the complex mult istage operations involved in HBsAg production
and purif ication, product yie ld is d ictated by the eff ic iency of the
detergent-mediated HBsAg liberation s tep while the purif ication
performance of membrane fi ltration and chromatographic operations
are influenced by the level of contaminants such as yeast proteins and
lipids . This thesis investigates the development of an improved primary
purif ication strategy to achieve higher HBsAg recovery and a cleaner
output to reduce the burden on downstream operations.
2.12 Pichia pastoris, a novel expression system
As a eukaryot ic organism, Pichia pastoris has many of the
advantages of higher eukaryotic express ion systems especial ly pos t-
translational modif ications such as protein process ing, protein folding
and protein secretion into the medium, along with the later facil itates of
easy puri f ication (Hollenberg and Gellissen, 1997). while being
manipulated as easy as Escherichia coli or S. cerevis iae. It is fas ter,
easier and less expens ive to use when compared to other expression
systems such as baculov irus or mammalian system, and generally
gives higher expression level. As yeast, it shares the advantages of
molecular and genetic manipulations with Saccharomyces, and i t has
added advantage of higher levels of expression. These features along
with easy maintenance, easy scale-up, and inexpensive growth
requirements make P. pastoris as a very useful protein expression
system. The process can be scaled up to a level of expression, which is
10-100 times higher than that of E. coli (Clare et a l., 1991; Faber et a l.,
Review of Literature
45
1995; Vozza et al., 1996). In many instances, heterologous proteins
were produced at much higher level when compared to their respective
productiv it ies in the tradit ional S. cerevisiae host. Their growth to high
cell density prov ides a means to produce large quantit ies of antigen
economically (Romanos et a l., 1992). Since 1984, over 300
heterologous proteins have been well expressed in P. pastoris (Cregg
et al., 2000; Cereghino et al., 2000). P. pastoris has gained more
popularity because of several factors such as alcohol oxidase I (AOX
I), which is one of the strongest and most regulated promoters, its
ability to integrate expression plasmids in its own genome in one or
more specif ic s ites, its ability to culture s trains in high density
fermenters; and its ready availabili ty as a kit ( Inv itrogen, USA).
Transformat ion of P. pastoris can easily be carried out either by
electroporation method or spheroplasting method, using linearized
recombinant plasmids and eff ic ienc ies are usually in several orders of
magnitude below those for other yeasts . Pretreatment of P. pastor is
with 0.1 M l i th ium acetate (LiAc) and 10 mM dithiothreito l (DTT) before
electroporation increased transformation eff ic iency approximately 150-
fold (Wu and Letchworth, 2004).
2.13 Detergent-mediated HBsAg liberation
The HBsAg product remains permanently localised on the yeast
endoplasmic reticulum (ER) fo llowing protein translation as transport
through the secretion pathway is blocked (Biemans et al. , 1992). A
detergent is required to facil i ta te the l iberation of HBsAg from tightly
Review of Literature
46
associated ER membrane components (Dekleva, 1999). Typically Triton
X-100, a non-ionic detergent, is employed for this purpose although
alternative detergents such as polysorbate 20 and 80 (PS 20 and PS
80) and variants of Triton such as Tri ton X-101, CHAPS (3-[(3-
Cholamidopropyl)dimethylammonio]-1-propanesulfonate) and glycol
ether solvents have been reported for s imilar appl ications (Wijnendaele
et a l. , 1987; Kniskern and Hagopian, 1997; Allen et al ., 2007). The
key feature to a detergent’s function is in its amphipathic structure
comprising a hydrophil ic “head” region and a hydrophobic “tai l” region
which can assoc iate with the hydrophobic surfaces of proteins.
Detergents fac ilitate the recovery of membrane proteins by disrupting
the bipolar l ipid membrane of cells and by forming protein-detergent
complexes which are soluble. Of the detergents commonly employed
for the recovery of l ipid-envelope VLPs, Triton and polysorbate, both
fall under the non-inonic detergent class . Their difference lies in the
polymer type and length of the hydrophobic “tails”. Triton detergents
are made from ethyleneglycoether polymers and have shorter
hydrocarbon chains and higher CMC values than polysorbate
detergents which contain oxyethylene polymers (Cal igur, 2008).
Detergents with lower CMC values are more stable however the
interaction between the detergent and the protein is also stronger
which could be an issue for subsequent detergent removal efforts.
CHAPS in contrast is a zwitter ionic detergent which may have the
added benefit of being more effective for disrupting protein-protein
bonds in reduc ing aggregation due to its ionic nature although hav ing a
Review of Literature
47
net neutral charge. In general, anionic or cationic detergents are less
preferred for biological applications as their charge charac teris tics
could modify protein s tructure leading to increased risks of protein
denaturation. Specif ically for HBsAg liberation, there are several
hypotheses as to how the detergent functions, mechanistically. Chi et
al (1994) proposed that detergents weaken the hydrophobic
interactions that ex ist between the VLP and the membrane
components. According to Sitr in and Kubek (1992), detergents facilitate
the disassociation of the HBsAg from the ER by promot ing the
separation of the yeast ER membrane from other unwanted cellular
debris. A study by Smith and co-workers suggests that detergent may
also promote VLP vesicle formation by inward budding of the ER
membrane (Smith et a l. , 2002). From the above, it appears that high
detergent concentrations would favour HBsAg liberation based on
reaction kinetics. However, to be taken also into cons ideration, is the
physiochemical nature of the HBsAg partic les which are 75% protein
and 25% lipid by mass (Gav ilanes et al. , 1982). There has been a lack
of consensus as to the optimal concentration of detergent for this
purpose. Other groups have reported that excess detergent could lead
to HBsAg partic le delipidation (Skelly et al. , 1981; Howard et a l. , 1982;
Sanchez et a l. , 1983) which in turn could result in a loss of antigenic
activ ity. This is unsurpris ing given that the lipid moiety has signif icant
roles in the maintenance of the structural and antigenic properties of
the protein components of the HBsAg (Gav ilanes et al., 1990).
Review of Literature
48
The amount of detergent could also have indirect implications on
the contamination prof ile of the result ing process stream. Large
amounts of l ipid are produced by yeast cells over extended periods of
cell culture (Wijnendaele et al., 1987) which in the case of HBsAg
production is necessary to achieve the desired product t itres. The
majority of these lipids are assoc iated with membranous organelles
which also contain some proportion of proteins for spec if ic metabol ic
functions . During the detergent step, it is l ikely that the membrane
associated lipids and proteins are co-released alongside the HBsAg
product and the extent of co-l iberation is also a function of detergent
concentration.
Based on the factors h ighlighted above, it is cr it ical to identi fy a
suitable detergent and its optimal range of operating concentrations for
effective recovery of HBsAg, preservation of its antigenic properties
and minimal co-release of contaminating host l ipids and proteins .
2.14 Precipitation for partial purification
Precipitation is a tradit ional protein purif ication method and is
examined here as a means of achiev ing partial purif ication and
enrichment of HBsAg prior to h igher resolution chromatographic
operations. Polyethylene glycol (PEG) and ammonium sulphate are the
most widely cited precipitating agents for VLPs and are assessed for
the primary purif ication of HBsAg based on the degree of contaminat ion
reduction and product yield.
Review of Literature
49
Polyethylene glycol (PEG) is attractive as a prec ipitating agent
owing to its low potential for protein denaturation even when present at
high concentrations (Ingham et a l. , 1990; Tsoka et a l. , 2000) and its
high fractionation eff ic iency when puri fying large size proteins such as
VLPs (Juckes et al., 1971). The appl ication of PEG precipitation for
HBsAg puri fication is not entirely new. Puri f ication of the f irst
generation hepati tis B vaccine based on HBsAg from plasma of carr iers
relied on mult iple PEG precipitation “cuts”. Since the recombinant
HBsAg examined here is not ident ical to the human form due to
differences in d isulphide c ross-linking, these recombinant partic les may
have different solubility behav iours in the presence of PEG. There is
also a difference in the role of the PEG prec ipitation step. In the 1970s
process for human HBsAg, PEG precipitation was the purif ication
workhorse and mult iple “cuts” were necessary to achieve the desired
f inal purity, often at the expense of s ignif icant product loss. As a
primary purif ication process for future-generation vaccines, its role
would be to reduce the contamination burden and process volume
entering downstream operations in a single “cut” with minimal product
loss . For PEG precipitation as a primary puri fication step product yield
is cr it ical and this rel ies on the effectiveness of precipitate recovery
from the liquor. Although PEG has a low intrinsic viscosity (Polson et
al. , 1964), concentrations above 10% w/v PEG may have substantial
effect on the v iscosi ty of the process stream. The ef fect of this on the
Review of Literature
50
performance of the centr ifugation or f i l tration step for the precipitate
recovery would need to be evaluated.
The use of ammonium sulphate, another popular precipitating
agent, is also investigated. It is generally known that the ammonium
sulphate salt is less selec tive compared to PEG which results in larger
quantit ies (> 40% w/v) of the salt being required for protein
precipitation (Wijnendaele et al., 1987) which could create waste
disposal issues. Lipid solubility, in contras t, is more sensit ive to
ammonium sulphate and lower concentrations of the salt are typically
suff ic ient for their precipitation.
Bracewell et a l., (2008) demonstrated an inverse purif ication
strategy us ing ammonium sulphate precip itation. In the study, l ipid
contaminants were precipitated out of solution whilst protein products
are reta ined in the soluble phase. This strategy would be more suitable
for large-scale VLP purif ication owing to the lower levels of ammonium
sulphate consumption which would pose less waste disposal issues.
2.15 Selective product release for primary purification
For expression of recombinant proteins in S. cerev isiae, there are
three main regions in the cell where the product of interest is l ikely to
be located: (1) the space between the cytoplasmic membrane and the
outermost mannan-protein layer of the cell wall , (2) in the cytosol, or
(3) embedded in subcellular organelles or partic les (Huang et al. ,
Review of Literature
51
1991). It is within the third category that the production of recombinant
l ipid-envelope VLPs such as the HBsAg falls . The natural local isation
of the product of interest in a specif ic region of the cell can provide a
powerful means of recovering the product separately from bulk
contaminants originating from the host cell. This approach has been
demonstrated prev iously in the recovery of proteins from the
periplasmic space of E.coli cel ls using osmot ic shock or enzymes
(French et al. , 1996; Witholt et al. , 1976) and from specif ic regions in
yeast using detergent and enzymes (Chi et a l. , 1994; Huang et al. ,
1991; Asenjo et al. , 1993). Using these methods, several-fold product
enrichment was reported with product yields of up to 90%. As
highlighted prev ious ly, HBsAg partic les in yeast are permanently
local ised on the endoplasmic reticulum (ER) because transport through
the secretion pathway is blocked (Biemans et al. , 1992). It is for this
reason that a detergent is typically added to the cel l lysate following
homogenisation to release the HBsAg from the assoc iated ER
components. The ability to separate ER partic les containing the HBsAg
product from bulk cel l contaminants prior to the addit ion of detergent
would be invaluable as this would allow the HBsAg product to be
recovered later and in a cleaner process stream. The endoplasmic
reticulum which is slightly hydrophobic (Chi et al. , 1994) and denser
than the cell cytosol would be sedimented together with solid debris
partic les when centrifuged whils t bulk cytosolic materia l would be in the
supernatant. By discarding the supernatant fraction and treating only
the sol ids fraction with detergent, HBsAg would be released into a
Review of Literature
52
process stream with s ignif icantly reduced host l ipid and protein
contaminants. The eff ic iency of this process would rely on the relative
partit ioning of the HBsAg product and host contaminants across the
cytosol and the membranous components.
It is possible to enhance further the selective recovery process by
f ine-tuning the cel l disruption condit ions. Characterisation of the size
distr ibution of yeast debris generated as a function of the
homogenisation condit ions revealed that the release of intracellular
proteins is often accompanied by fragmentation and micronisation of
the cell wall (Siddiqi et a l., 1995). Condit ions of the homogenisation
step could have a major influence on the eff ic iency of the selective
recovery process at three different levels: (1) the ex tent of cell
disruption would influence the accessibil ity of detergent to the ER
during the HBsAg liberation step and this would impact product yield,
(2) the population size of the ER components following homogenisation
would influence the effec tiveness of their recovery by sedimentation
during the init ia l centr ifugation s tep and this would also impact product
yie ld, and (3) the partic le size dis tribution of membranous debris would
influence the rate of co-ex trac tion of contaminating lipids and proteins
during the detergent step and this would impact the level of purif ication
achieved. Thus, a trade-off study on the impact of homogenisation
conditions would be warranted in developing any approach to selective
product release.
Review of Literature
53
2.16 Scalability of selective product release
In process development, USD techniques are invaluable as they
are capable of predicting industrial scale performance of
biopharmaceutical protein downstream process ing in a rapid fashion
using as lit t le as mil l i l itres of material (Titchener-Hooker et a l. , 2008).
Identif ication of process challenges and predic tion of performance of
large scale protein precipitation operations using ultra scale-down
techniques has been demonstrated (Boychyn et a l. , 2000). Hence, the
focus here was on the scale-up of a selec tive product release method.
The selec tive product release methodology investigated in this
work relies on a twin centrifugation operation. The role of the f irst
centr ifugation step is to allow the elimination of bulk cytosolic
contaminants whi lst retaining the solids fraction containing the HBsAg
product. The second centr ifugation step which is performed following
detergent treatment recovers the HBsAg product in the supernatant
whilst solid contaminants are discarded. Different centr ifuge designs
such as the tubular bowl, multichamber bowl and disc-s tack centr ifuge
have been rev iewed (Salte et al., 2006; Boychyn et a l. , 2004). The
CARR PowerfugeTM centr ifuge, which is a tubular bowl variant, is most
suited for inclusion in the selective product release process sequence
due to its good solids compaction and high c lar if ication features . The
level of dewatering and clar if ication attained would dictate the
selec tiv ity and product yield from this approach.
Review of Literature
54
The Sigma concept (Ambler, 1959) allows the comparison and
prediction of centr ifuge performance across scales and centr ifuge
designs by the introduction of correc tion factors which account for
differences in f low pat terns. USD models have been developed from the
Sigma theory (Tustian et a l., 2008; Boychyn et a l. , 2004; Boychyn et
al. , 2000; Salte et a l. , 2006) and are useful here to predict the
clar if ication performance of an industr ial CARR Powerfuge T M operating
at di fferent feed flow rates and rotational speeds. Hav ing determined
the optimal operating condit ions of the CARR Powerfuge T M centr ifuge
for the selective recovery process, a pilot-scale validation s tudy can be
performed to verify that the purif ication potentials of any methodology
are not compromised upon scale-up.