Potential conflict of interest: Dr. Locarnini is a consultant for and received grants from Gilead. He is a consultant for Bristol-Myers Squibb and Evivar Pty Ltd. He is also a consultant for and owns stock in Gilead. He received grants from LG Sciences. He holds intellectual property rights for Melbourne Health.
The hepatitis B virus (HBV) mutation that encodes rtA181T is selected in the viral polymerase during antiviral drug therapy and can also encode a stop codon in the overlapping surface gene at amino acid 172 (sW172*) resulting in truncation of the last 55 amino acids of the C-terminal hydrophobic region of the surface proteins. This mutation is usually detected as a mixed population with wild-type HBV. In vitro analysis revealed that the rtA181T/sW172* variant is not only defective in secretion of viral particles causing intracellular retention of surface proteins, it also has a dominant negative effect on virion but not subviral particle secretion when coexpressed with the wild type. This dominant negative effect was attributed to the truncated S protein alone. Furthermore, these truncated surface proteins were less glycosylated, and the truncated L protein was able to support virion secretion. Examination of sequential HBV DNA levels in patients failing lamivudine or adefovir therapy where only the rtA181T change was detected via polymerase chain reaction sequencing revealed that viral load rebound did not occur or was not as large as usually observed with drug-resistant HBV. Conclusion: The rtA181T/sW172* variant has a secretory defect and exerts a dominant negative effect on wild-type HBV virion secretion. The selection of rtA181T/sW172* reduced the typical extent of virological breakthrough, resulting in a missed diagnosis of drug resistance if viral load was used as the only criterion for drug failure, necessitating HBV polymerase chain reaction sequencing or other genotypic methods to diagnose antiviral drug resistance in these cases. (HEPATOLOGY 2008.)
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
Treatments for hepatitis B virus (HBV) infection include interferon and nucleos(t)ide analogs (NAs). However, treatment with NAs is hampered by selection of changes in the HBV reverse transcriptase (rt) associated with resistance. The region of the HBV genome that encodes the rt domains of the polymerase overlaps completely with the envelope; therefore, changes associated with NA therapy can cause the selection of envelope changes.1
HBV encodes three membrane-associated envelope proteins, all sharing the same stop codon, with translation initiating from different start codons. The smallest of these proteins, S, is encoded by the S gene, which produces a protein of 226 amino acids (aa). The middle protein, M, contains a further N-terminal extension of 55 aa encoded by the upstream Pre-S2 gene. The L protein has yet another N-terminal extension of 108-119 aa, depending on the genotype, encoded by Pre-S1. The major functions of the HBV surface proteins include envelopment of nucleocapsids with subsequent assembly of virions, and assembly into empty subviral particles that are secreted in great excess over HBV virions collectively referred to as hepatitis B surface antigen. The surface proteins also contain the major antigenic epitopes and are involved in hepatocyte binding and entry during infection.
The nucleotide mutation that encodes rtA181T has now been described in lamivudine (LMV), adefovir (ADV),2–4 telbivudine (www.accessdata.fda.gov/scripts/cder/drugsatfda/index.cfm), and clevudine5 therapy at varying frequencies. In vitro drug susceptibility testing has shown rtA181T confers low level resistance to a number of NAs.6 This mutation can also encode a stop codon in the overlapping envelope proteins at aa 172 in the S region (sW172*) resulting in truncation and loss of a large portion of the C-terminal hydrophobic region (Fig. 1). Several studies have demonstrated varying effects of C-terminal envelope protein truncations. A preliminary phenotypic analysis of rtA181T/sW172* by Yeh et al.4 revealed that rtA181T conferred resistance to LMV and that secretion of hepatitis B surface antigen and HBV DNA was impaired. Studies using the S protein alone have shown that similar truncations of the C-terminal hydrophobic region can inhibit hepatitis B surface antigen secretion7 as well as hepatitis delta virus secretion.8
In this study, we determined the frequency of the rtA181T/sW172* mutation and examined its effect on HBV replication. We demonstrated that rtA181T/sW172* is not only defective in secretion, but has a dominant negative effect on virion secretion in a mixed infection, which may affect the subsequent profile of virological breakthrough and the diagnosis of resistance if the serum HBV DNA level is the sole criterion.
The SeqHepB sequence analysis program and relational database9, 10 was used to determine the frequency of the rtA181T/sW172* mutation and correlating viral load and serum alanine aminotransferase data if available. At the time of the study, SeqHepB contained HBV polymerase sequences from 2,149 isolates from 1,440 chronic hepatitis B patients that had been submitted to VIDRL for drug-resistance testing via sequence analysis.
Cell Culture and Transfection.
Huh7 cells were maintained and transfected as described.10
HBV Expression Vectors.
The genotype D, 1.3 times genome length HBV construct was used (pHBVwt [wt, wild-type]).10, 11 A point mutation encoding both rtA181T and sW172 was introduced via site-directed mutagenesis (Stratagene, La Jolla, CA) using primer sW172* (Table 1) to generate pHBV172* (Fig. 2).
Table 1. Primers Used in This Study
Abbreviation: SDM, site-directed mutagenesis.
The gene for each of the surface proteins was polymerase chain reaction–amplified from pHBVwt using appropriate primers (Table 1) and cloned into the Not1 and Sal1 restriction sites of the pCI vector (Promega, WI) using standard protocols to generate plasmids pL, pM, and pS, which express the wild-type surface proteins designated L, M, and S, respectively. For pL, the start codons for M and S were mutagenized, while for pM the start codon for S was mutagenized to prevent internal initiation of translation using the primers listed in Table 1. Each of these clones were also mutagenized to create a stop codon at aa S172 (sW172*) as described above to generate pLt, pMt, and pSt, which express truncated surface proteins Lt, Mt, and St, respectively. A stop codon at position s152 was also introduced into pLwt (pL152*) to serve as a control for the trypsin protection assay (Fig. 2).
Western Blot Analysis.
For intracellular proteins, cells were lysed in 1% Nonidet P40 (NP40), 2% sodium dodecyl sulfate, 150 mM NaCl, 50 mM trishydroxymethylaminomethane-HCl (pH 7.5), and protease inhibitor cocktail (Roche Diagnostics, Mannheim, Germany). For extracellular proteins, cell culture supernatants were used directly. Samples were mixed with Laemmli buffer (10% sodium dodecyl sulfate; 40% glycerol; 0.05% bromophenol blue; 25% β-mercaptoethanol; 0.25 M trishydroxymethylaminomethane [pH 6.8]), boiled, separated via sodium dodecyl sulfate–polyacrylamide gel electrophoresis, and transferred onto nitrocellulose membranes according to standard protocols. Membranes were blocked overnight in 3% skim milk powder in phosphate-buffered saline–Tween-20 (0.1%). Surface proteins were detected using antibody H16612 (kindly donated by Paul Coleman, Abbott Laboratories, Abbott Park, IL). For specific detection of L, anti-PreS1 (Biogenesis, Poole, UK) was used. Corresponding secondary antibodies labeled with horseradish peroxidase were used for detection via enhanced chemiluminescence using X-ray film.
Southern Blot Analysis.
HBV DNA was analyzed via Southern blotting using methods described previously.10, 11 Virions were immunoprecipitated from the supernatant prior to Southern blotting to exclude unenveloped nucleocapsids, as described in the Immune Precipitation section below.
For each sample, 4 μg of anti-S antibodies HBV-Ab-17 and HBV-Ab-1913 (kindly donated by Rachel Eren, XTL Bio, Rehovot, Israel) were mixed together and bound to 50 μL of 50% Protein A Sepharose-CL4B (GE Healthcare, Chalfont St. Giles, UK) for 24 hours at 4°C, rotating. Beads were washed several times and incubated overnight with supernatant at 4°C, rotating. The beads were then washed three times with PBS, the sample was dissolved in 0.5% NP40 lysis buffer, and HBV DNA was purified as described in the Southern Blot Analysis section above.
Transfected Huh7 cells were trypsinized at 3 days after transfection and reseeded into dishes containing glass coverslips. After a further 2 days' growth, cells were washed with PBS, fixed in 100% methanol for 15 minutes, and incubated with an anti-S monoclonal antibody (H53, Abbott Laboratories) in PBS at 37°C in a humidified incubator. The primary antibody was detected using the MACH3 mouse probe HRP-polymer kit (Biocare Medical, Concord, CA) and the cell nuclei stained with hematoxylin.
Trypsin Protection Assay.
A trypsin protection assay was used to determine the topology of the Pre-S regions of Lt, as described elsewhere.14
Particles from the supernatants of transfected Huh7 cells were pelleted through 4 mL 20% (wt/wt) sucrose cushion in trishydroxymethylaminomethane-buffered saline at 96 000 g in an SW28 rotor for 6 hours at 10°C. Pellets were washed, resuspended in Dulbecco's modified Eagle's medium, and examined with immunosorbent electron microscopy as follows: 300-mesh carbon-coated grids (ProSciTech, Kirwan, Queensland, Australia) were precoated with an anti-HBs monoclonal antibody (HBV-Ab 19, XTL Bio, Rehovot, Israel) for 1 hour, then washed with PBS. A drop of the inocula was then added to the antibody-coated grid and incubated for 2 minutes. Grids were washed three times with PBS before drying, stained with 3% phosphotungstic acid (pH 7.4), and viewed using a Tecnai 12 electron microscope (FEI, Hillsboro, OR) at 120 kV.
A review of the 1,440 patients in the SeqHepB database revealed that HBV encoding rtA181T/sW172* was detected by itself in 1%, and in association with M204I/V in 1% of LMV-resistant cases (2% of total LMV-resistant cases). In contrast, rtA181T/sW172* was detected alone in 11%, and in combination with rtN236T in 7% of ADV-resistant cases (18% of total ADV-resistant cases) (Table 2). In patients treated with LMV followed by LMV/ADV combination therapy, rtA181T/sW172* was detected in combination with rtM204I in 42% of cases. HBV encoding rtA181T/sW172* was observed in only 0.2% of isolates from untreated patients or where the specific therapy at the time of specimen submission was not known (Table 2). rtA181T was detected as a mixed population (rtA181A/T) with wild-type in approximately 80% of these isolates, and clonal sequence analysis from one of these samples revealed that rtA181T was present in 53% of clones (data not shown). rtA181T/sW172* was detected across all HBV genotypes (data not shown).
Table 2. Clinical Occurrence of rtA181T and Other Resistance Mutations and Corresponding Surface Mutations
Reverse Transcriptase Mutation
Results are shown as number of patients followed by the percentage of that mutation in total patients in the cohort in parentheses.
LMV (n = 560)
A181T + M204I/V
W172* + l195M/W196S/L
ADV (n = 44)
A181T + N236T
A181V + N236T
LMV + add-on ADV (n = 7)
A181T + M204I
W172* + I196S/L
No reported therapy (n = 830)
Sequential analysis of serum HBV DNA from five patients on NA therapy in which rtA181T/sW172* was the only drug-resistant mutation detected by sequencing revealed an atypical viral load pattern. In two cases (LMV therapy), there was only an incremental increase in HBV DNA levels of between 1 and 2 log10 over 12 months. In the remaining three cases (1 LMV therapy, 2 ADV therapy), there was essentially no change in HBV DNA levels over 12 months. rtA181T/sW172* was detected at nadir in all five patients, which was approximately 105-106 copies/mL. Alanine aminotransferase data of these patients were incomplete, but where available showed either no change or increased levels (data not shown).
HBV DNA Replication.
The HBV172* mutant had less intracellular replicative intermediates than HBVwt (Fig. 3A; Fig. 4A, column 1 versus 5). Extracellularly, no HBV DNA was detected via Southern hybridization (Fig. 3B; Fig. 4B, column 5).
HBV Surface Protein Glycosylation and Secretion.
A truncation of 55 aa that occurs with sW172* results in a decrease in molecular weight of approximately 6 kDa. This was observed on western blot analysis, whereby the nonglycosylated forms of the truncated proteins showed bands at approximately 18 kDa (St) and 33 kDa (Lt) (Fig. 3). Mt was barely detectable in this study, but in these experiments the levels of Mwt are very low, which is a property consistent with this construct and cell line (unpublished observations).
Intracellularly, St was present at similar levels to S, but had much lower levels of glycosylation. In contrast, the intracellular levels of Mt and Lt were greatly diminished compared with M and L and had less glycosylation. Decreased glycosylation of Lt was also observed when expressed alone (see Trypsin Protection Assay). These truncated surface proteins were only detectable intracellularly and were not secreted into the supernatant (Figs. 3 and 4).
Effect of Coexpression of HBVwt and HBV172* on Secretion.
To determine whether the coexistence observed in vivo is due to a rescue of the secretion defect of HBV172* by HBVwt, secretion of virion DNA and surface proteins was measured from cells transfected with mixtures of pHBVwt and pHBV172*.
Examination of HBV DNA revealed that intracellular core DNA was detectable in all samples tested, but was slightly less for HBV172* alone (Fig. 4A). When secreted virions were measured, maximal secretion occurred from cells transfected with HBVwt alone, whereas no virions were secreted from cells transfected with HBV172* alone. However, when HBVwt was present in four-fold excess over HBV172*, virion secretion was greatly decreased compared with HBVwt. When both constructs were transfected in equal amounts, virion secretion was barely detected. Virion secretion was not detected at all when HBV172* was expressed in four-fold excess over HBVwt (Fig. 4B). This negative effect on secretion was not due to simple dilution of HBVwt (Fig. 4D).
When expressed with HBVwt, the truncated surface proteins exhibited an unexpected banding pattern. The intracellular levels of St increased as the level of transfected pHBV172* DNA increased, whereas the intracellular level of Lt did not increase as input plasmid DNA increased; rather, the accumulation of Lt was dependent on expression with wild-type, the maximal accumulation occurring at a ratio of 1:1 (Fig. 5A). The truncated surface proteins were also secreted into the supernatant when wild-type proteins were present (Fig. 5B). Lt was confirmed to be Lt and not M by stripping the blot and reprobing with an antibody specific for Pre-S1 (Fig. 5C).
The intracellular localization of surface proteins in Huh7 cells transfected with pHBVwt, pHBV172*, or a 1:1 mixture of both was examined via immunohistochemistry. The surface proteins from HBVwt exhibited granular, packeted cytoplasmic staining (Fig. 6A). The surface proteins from HBV172* exhibited very fine granules that were not packeted, and were more diffuse and streaky, and also had substantially greater staining intensity than HBVwt, consistent with intracellular retention (Fig. 6B). This accumulation was also observed when cells were stained with the H166 antibody (which recognizes an epitope that is linear and conformational), and detected with immunofluorescence (data not shown). When equal amounts of both HBVwt and HBV172* were transfected together into Huh7 cells, a blend of the two phenotypes was observed wherein the staining pattern was a mixture of granules and streakiness and the intensity of staining was greater than wild-type, though not as marked as HBV172*; nevertheless, these findings are consistent with intracellular retention (Fig. 6C).
Topology of Lt.
To determine the topology of the PreS regions of Lt, microsomal vesicles from Huh7 cells transfected with pL, pLt, or pL152* were prepared and subjected to a trypsin protection assay. There are several trypsin cleavage sites in the PreS regions of L and M, the most distal located at aaR47 in M (Fig. 1). Cleavage only occurs when the PreS regions are localized to the surface of microsomal vesicles, which corresponds to a cytosolic localization intracellularly. The PreS region of L truncated at aa152 is unable to translocate,15, 16 and thus served as a control. All full-length and truncated L proteins from vesicles that were not treated with trypsin had the expected banding patterns (Fig. 7) and were degraded when treated with both trypsin and NP40 as expected. When intact vesicles were treated with trypsin, L152* was completely cleaved by trypsin (control), whereas L and Lt were partially protected from cleavage (Fig. 7). These results demonstrate that the PreS region of Lt is able to translocate and hence can form the same topology as L.
Effect of S and St on Virion Secretion When Expressed with pHBVwt or pHBV172*.
To determine whether the dominant negative effect of HBV172* on virion secretion could be attributed specifically to the truncated S protein, cells were transfected with pHBVwt alone or cotransfected with pS or pSt, and virion secretion was measured via immunoprecipitation and Southern blotting. Coexpression of St with HBVwt caused a reduction in virion secretion, whereas coexpression of HBVwt with S did not decrease virion secretion (Fig. 8B).
pHBV172* was also transfected alone or was cotransfected with pS or pSt into Huh7 cells, and virion secretion was measured. When pHBV172* was cotransfected with pS, HBV virions were secreted at very low levels, whereas St was unable to restore virion secretion (Fig. 8D). These data demonstrate not only that S alone is sufficient to rescue secretion of viral particles from HBV172*, but that Lt is capable of supporting virion secretion.
The morphology of particles secreted from Huh7 cells transfected with pHBVwt or equal amounts of pHBVwt and pHBV172* was examined via electron microscopy (EM). Particles from HBVwt comprised virions as well as an excess of empty 22-24nm spherical and filamentous particles, as expected for HBV. In contrast, particles secreted from cells transfected with a mixture of both HBVwt and HBV172* contained fewer 42-nm virions than HBVwt, confirming the Southern blot results, and most of the subviral particles were spherical, smaller, and of dimpled appearance. Filaments were very rare, and when seen were much shorter than those observed for HBVwt (Fig. 9).
Treatment of patients with chronic hepatitis B using ADV, LMV, clevudine, or telbivudine can result in the emergence of HBV encoding rtA181T/sW172*, typically as a mixed population with wild-type HBV (rtA181A/T). In this study, patients who selected only rtA181T/sW172* during therapy did not exhibit typical virological breakthrough profiles. Extensive in vitro analysis of this variant revealed that the rtA181T/sW172* mutant is not only defective in secretion, but has a dominant negative effect on wild-type virion secretion. Direct experimental evidence for the dominant negative effect of rtA181T/sW172* was provided using several approaches, including Southern blotting, immunohistochemistry, and EM. The addition of even small amounts of rtA181T/sW172* to HBVwt can result in a dramatic inhibition of secretion of virions (Fig. 4B, lanes 2, 3, and 4). Furthermore, the consequence of this dominant negative effect is shown in Fig. 6, where dual transfection and replication results in intracellular retention. Finally, no virions were detected in the supernatant of these cells using EM; instead, an abundance of subviral particles secreted from the cell was observed (Fig. 9). This dominant negative effect on virion secretion was attributed to the truncated S protein alone (Fig. 8B, lanes 5 and 6). Other important observations included that these truncated surface proteins were less glycosylated than their wild-type counterparts, and that the truncated L protein was able to support virion secretion. Together these data provide strong and compelling evidence for the dominant negative effect of rtA181T/sW172*.
The relevance of the rtA181T/sW172* mutation in the clinical situation may be more than reduced sensitivity to antivirals.6, 17 The sequence of events as viral resistance to LMV (rtL180M + rtM204I/V), LdT (rtM204I), or ADV (rtN236T) starts with the detection of mutations in the reverse transcriptase, then steady increase in serum HBV DNA levels, and as a very late effect, a rise in serum aminotransferases.18–21 In this study, we demonstrated that rtA181T/sW172* has a secretory defect and exerts a dominant negative effect on wild-type HBV virion secretion. Thus, the selection of rtA181T/sW172* can affect the diagnosis of resistance if only viral load is used. At least a 1.0 log10 IU/mL increase of HBV DNA over 3-6 months from nadir is usually observed with LMV19 or ADV,22 and the current case definition of antiviral resistance is defined as a >1.0 log10 IU/mL increase in serum HBV DNA level from nadir in two consecutive samples taken 1 month apart.23 In those five clinical cases where only rtA181T/sW172* was detected, the viral load rebound was not as large as is typically observed during viral breakthrough. This phenomenon has also been described by others.6 Thus, the selection and emergence of rtA181T/sW172* will mask the diagnosis of resistance if only serum HBV DNA levels are used. In those patients being treated with NAs such as LMV, telbivudine, ADV, or clevudine that select out rtA181T/sW172*, HBV polymerase sequencing or other genotypic methods such as the line probe assay will be needed to diagnose antiviral drug-resistance.23 It is important to note that rtA181T/sW172* was present in all five cases at nadir, which ranged from 105-106 copies/mL, a level at which population-based polymerase chain reaction sequencing methods are reliable.24
The wild-type HBV surface proteins are synthesized in the endoplasmic reticulum (ER) where they rapidly undergo dimer and multimer formation via extensive disulphide bonding, eventually resulting in budding into a post-ER, pre-golgi intermediate compartment as either spherical or filamentous empty subviral particles, or as virions if nucleocapsids from the cytoplasm bud into these membranes, a process that requires both L and S, as well as host factors. The S protein alone can be secreted as subviral particles, whereas secretion of L is dependent on S. Particles can comprise any combination of glycosylated and nonglycosylated L, M, and S proteins, depending on the abundance of the different glycoforms in the ER membrane.25 The data presented here show that the truncated surface proteins are not secreted via this pathway, but are retained and accumulate within the cell; however, we have not demonstrated where the block in the pathway occurs.
These truncated proteins can be secreted when coexpressed with HBVwt or S alone, although virion formation is restricted and there is still some intracellular accumulation. This is likely due to disulphide bonding of truncated and wild-type surface proteins and subsequent budding into subviral particles containing both truncated and full-length surface proteins that are secreted in the same manner as particles containing only wild-type surface proteins; however, presence of the truncated surface proteins still has some inhibitory effect. EM was used to show that the particles secreted in a mixed infection were predominantly spherical subviral particles, with very few virions and very few relatively short filaments. The decrease in filaments may be due to defective or low levels of L, as L abundant in filamentous, not spherical subviral particles.26–28 The accumulation and secretion of Lt observed in the presence of wild-type (Fig. 5) is consistent with disulphide bonding of the relatively unstable Lt with wild-type surface proteins, which may result in partial protection from degradation by stabilizing the structure of Lt and budding into viral particles.
When the surface proteins in cells transfected with pHBV172* were analyzed via western blotting, Lt and Mt were present at very low levels, whereas St was abundant. The reason for this difference is probably posttranscriptional, because there was no difference in RNA levels observed with northern blotting (data not shown). One possible explanation for the decrease in Lt and Mt is targeted proteasomal degradation of these forms. By observing their accumulation when transfected cells were treated with the proteasome inhibitor lactacystin, we found that St was degraded slightly more than Swt, but the most degradation occurred for Mt and Lt (unpublished observations). In a study where HBV-expressing cells were treated with glucosidase inhibitors (which specifically affect glycosylated proteins), the glycosylated and nonglycosylated forms of L and M were degraded equally, suggesting that proteasomal degradation occurs after the glycosylated proteins have formed complexes with unglycosylated proteins, possibly in the form of virions or subviral particles.29 Accordingly, one possible explanation for the dominant negative effect on virion secretion of HBV172* is that the truncated surface proteins interact with wild-type surface proteins and could be assembled into virions. The truncated surface proteins present in the envelope of some of these virions could then be targeted for proteasomal degradation, taking with it the entire virion. However, subviral particle secretion is not detectably affected, consistent with studies showing that subviral particles and virions are formed by different mechanisms.30, 31 The accumulation of Lt in the presence of HBVwt (Fig. 4A) and its ability to support virion secretion (Fig. 8D) is consistent with this hypothesis.
Another possible explanation for the dominant negative affect on virion secretion is the decreased glycosylation observed with the truncated surface proteins. For glycosylation to occur, aa sN146 must be located on the lumenal side of the ER (Fig. 1), although only around half of wild-type surface proteins are glycosylated despite them all having the same topology at the glycosylation site.25 The glycosylation machinery is required for HBV secretion, but whether direct glycosylation of the surface proteins or glycosylation of a cellular protein is required remains open.32–37 If glycosylation at the correct ratio is indeed important for viral particle secretion, the altered glycosylation observed here may be the cause of the secretion defect. Alternatively, the presence of the secretion-defective truncated surface proteins together with wild-type may affect secretion by numerous mechanisms not discussed here, including hindering interactions with host cell machinery required for secretion and the lower intracellular levels of L.
In summary, we have shown that rtA181T/sW172* is dependent on wild-type HBV for secretion, but acts as a dominant negative mutant for HBV virion secretion resulting in retention within the cell. Thus, due to this secretory defect, selection of rtA181T/sW172* can result in a nonclassical viral rebound, which could result in a missed diagnosis of drug resistance if only viral load is used as the sole criterion for diagnosing viral breakthrough.18, 23 Finally, based on these findings, it may become necessary to monitor patients on NA therapy more frequently for the emergence and selection of rtA181T/sW172* by sequencing the HBV polymerase using standard polymerase chain reaction approaches or other genotypic methods,23 especially when the viral loads drop slowly or plateau early and there is a clinical suspicion of antiviral drug failure.
The authors thank Dr. Michelle Yong and Dr. Lilly Yuen (VIDRL) for help with collating clinical data; Lucy Selleck (VIDRL), Peter Revill (VIDRL), and Lynne Waddington (CSIRO) for technical assistance; Dr. John Pedersen (TissuPath Laboratories) for assistance with immunohistochemistry; Dr. Paul Coleman (Abbott Laboratories) and Dr. Rachel Eren (XTL Bio) for supplying antibodies; and Professor Chloe Thio (Johns Hopkins University) for reading the manuscript.