A comprehensive system for consistent numbering of HCV sequences, proteins and epitopes

Abstract

In October 2004, an expert meeting was convened in parallel with the 11th Symposium on Hepatitis C and Related Viruses to discuss how HCV sequence databases could introduce and facilitate a standardized numbering system for HCV nucleotides, proteins and epitopes. Inconsistent and inaccurate numbering of locations in DNA and protein sequences is a problem in the HCV scientific literature. Consistency in numbering is increasingly required for functional and clinical studies of HCV. For example, an unambiguous method for referring to amino acid substitutions at specific positions in NS3 and NS5B coding sequences associated with resistance to specific HCV inhibitors is essential in the investigation of antiviral treatment. This article provides a practical guide to help circumvent these problems in the future, and to bring a common language into discussions in the field. The scope of the current system is limited to the HCV polyprotein and the untranslated regions (UTRs); because of the controversial nature and extreme length variation of the alternate reading frame proteins, numbering for these proteins, if needed, will be decided at a later date. We propose a numbering system adapted from the Los Alamos HIV database,1 with elements from the hepatitis B virus numbering system.2 The system comprises both nucleotides and amino acid sequences and epitopes. It uses the full length genome sequence of isolate H77 (accession number AF009606) as a reference, and includes a method for numbering insertions and deletions relative to this reference sequence. H77 was chosen because it is a commonly used reference strain for many different kinds of functional studies. Furthermore, RNA transcripts from this sequence are of demonstrated infectivity,3, 4 providing evidence that the 5′ and 3′ ends of the sequence are complete. Table 1 lists the boundaries of HCV genomic regions and Fig. 1 provides detailed nucleic acid and amino acid numbering over the complete AF009606 HCV genome sequence. The complete AF009606 HCV genome sequence. HCV, hepatitis C virus; UTR, untranslated region; PPT, poly-pyrimidine tract; CDR, complementarity-determining region. Numbering an amino acid sequence can be absolute, i.e., based on the HCV polyprotein with numbering starting at the first amino acid of the core protein and continuing through the end of NS5B; or it can be relative, i.e., starting over at every protein (core is numbered from 1 to 191, then E1 from 1 to 192). Both of these numbering systems are used in HCV research. While most epitopes are numbered following the polyprotein numbering, drug resistance mutations tend to be numbered using protein numbering, following the practice for HIV-1 and now HBV. All HCV sequence databases incorporate both numbering systems. In order to avoid any confusion between the absolute and the relative numbering, the protein coordinates should be preceded by the protein name when using the relative numbering (e.g., NS3-A156T). Conversion between the relative and absolute numbering can be easily done using either the Locator tool, available at the American HCV database website (http://hcv.lanl.gov/content/hcv-db/LOCATE/locate.html), or the Number tool on the European HCV database website (http://euhcvdb.ibcp.fr/euHCVdb/). There are two different methods for the numbering of the 5′ UTR that each have their adherents and opponents. One system starts numbering the polyprotein coding sequence at 1, and continues on through the end of the 3′ UTR; the 5′ UTR is numbered in the reverse direction, starting at −1 and continuing on to −341. The other system simply starts at 1 for the first nucleotide of the 5′ UTR, and continues on to the last nucleotide of the 3′ UTR. Both systems have advantages and disadvantages, and one can easily be converted to the other. For reasons of simplicity we have decided to adopt the numbering system that sets the start of the 5′ UTR at 1. Nucleotide numbering continues uninterrupted into the coding sequence. In the case of AF009606, the AUG initiation codon would be numbered as 342–344. This system avoids the practical problem encountered in most sequence editors of numbering a sequence non-consecutively (i.e., in the negative numbering system for the 5′ UTR, the numbering at the 5′ UTR/core junction should proceed …−3, −2, −1, +1, +2, +3…, whereas sequence editors tend to number the −1 base as zero). Numbering of the 3′ end of the HCV genome has to accommodate a stretch of pyrimidine residues of highly variable length between the 40 alignable nucleotides at the start of the 3′ UTR and the end of the 3′ UTR, including the highly conserved 3′X-tail. Diagrammatically, the 3′ UTR comprises the following sequential stretches (AF009606 numbering): Coding sequence […] CTC CCC AAC CGA TGA - 9377 3′UTR start (variable region): 9378–AGGTTGGGGTAAACACTCCGGCCTCTTAG GCCATTTCCTG - 9417 Spacer region: 9418 - <60–100 pyrimidines, mainly T> -9548 3′UTR end (conserved region or 3′X): 9549–GGTGGCTCCATCTTAGCCCTAGTCACGGCTAGCTGT GAAAGGTCCGTGAGCCGCATGACTGCAGAGAGTGCTGATA CTGGCCTCTCTGCAGATCATGT - 9646 The length of the poly-pyrimidine tract (PPT) changes rapidly over time within an experimentally infected chimpanzee5 and is known to vary between different clones of the original H77 isolate3, 4 (accessions AF009069-77). The actual numbering of the 3′ end of the 3′ UTR sequence beyond the PPT is therefore of no significance. Furthermore, since the alignment of the PPT is arbitrary, so are any designation of insertions or deletions and any numbering attempt in this region. An additional complication is caused by the choice of H77 as the reference sequence. The 3′X region of H77 has AAT as the first 3 nucleotides. However, this is not the case in most other isolates in which the 3′X is only 98 nucleotides long. To prevent complications arising from this unusual feature of H77, for numbering purposes, the AAT of H77 should be considered as part of the PPT rather than the 3′X region. With this exception, the 3′ UTR follows the numbering of the reference sequence AF009606. If the sequence under consideration has a PPT that is not longer than AF009606 (no sequence currently in the HCV database has a longer PPT), the numbering is straightforward. For a sequence with a longer PPT, we propose that the insertion is ignored for the purpose of numbering. This means that the first nucleotide of the continuation of the alignment after the PPT, the region described in3) and numbered 9418 in their paper, will be numbered 9549, regardless of whether this is the actual position. Positions after that will again be numbered consecutively. Mutations are numbered according to their position, e.g., a NS5A:R217K would be used to indicate that the Arginine in position 217 of the NS5A protein has mutated to a Lysine. The codon involved changes from AGG to AAG (numbering 6906–6908), so the corresponding nucleotide mutation would be denoted G6907A. Insertions can be incorporated by basing the numbering on an alignment that includes all possible insertions, or by devising a special numbering system for insertions relative to a given numbering system. There are several reasons why the second method is preferable. First, sequences occasionally contain very long insertions, meaning that it would be hard to define an alignment that could accommodate all future insertions, and changing the numbering system in a way that would invalidate all previous numbering would be very disruptive. This problem is circumvented by the method outlined below. Second, insertions in HCV are relatively rare, so that this notation does not have to be used often. Deletions are much simpler to deal with than insertions, and we outline a simple naming method for unequivocally designating those. 1. Insertion in sequence relative to the reference sequence. For these cases we propose a residue number/alphabet, where inserted bases or amino acids are indicated by lower case letters following the nucleotide or amino acid position where they occur. For example, three inserted bases in an HCV variant inserted between positions 131 and 132 in the AF009606 reference sequence would be described as 131a, 131b and 131c. (For insertions longer than the length of the alphabet, numbering would proceed 131x, 131y, 131z, 131aa, 131ab, 131ac, …131az, 131ba, 131bb, 131bc…). A similar scheme has been used for numbering amino acids in the immunoglobulin complementarity-determining region (CDR) loops (e.g.,6). Example: in the following NS5A fragment, the location in the variant of the inserted aspartate (D) between AF009606 residues 2412 and 2413 would be referred to as D2412a or NS5A-D440a. 2411 2415 amino acids from start of H77 polyprotein | | GA-DTE - AF009606 EADDTE - variant sequence Table 2 shows the numbering of each variant amino acid in this example. 2. Deletion in sequence relative to the reference sequence. Mentioning the deletions can be useful in cases where the length of the fragment is important, or when referring to the location of an amino acid that is located after a gap (relative to the reference sequence) in an epitope. Example: in the following NS5A fragment, the variant region would be numbered 2410–2415 (del 2413) or NS5A-438-443 (del 441) to make this explicit. 2410 2415 H77 AA position from start of polyprotein | | SGADTE - AF009606 EEA-TE - variant Table 3 shows the numbering of each variant amino acid in this example. Mentioning the deletions would be useful in cases where the length of the fragment is important; in this example, if the deletion were not made explicit it could be assumed that the range 2410–2415 refers to a peptide that is 6 AA long, when it really is only 5 AA. 3. Synthetic sequences. A separate problem is that of artificially engineered sequences, such as replacement of stretches of the HCV genome with extraneous genetic segments. The numbering system should be able to cope with this situation also, by treating the extraneous segment as a special type of HCV, which can be numbered according to the length of the insert. Finding the stretches that correspond to “real” HCV depends on pattern matching, which requires accurate location of the end and re-start of HCV sequences on either side of the insert to retain correct numbering. Using the same example, 2411 2415 amino acids from polyprotein start | | | GADT——-E - AF009606 GADTYNTVATLE - variant sequence ======= insert Table 4 shows the numbering of the insert in this example. While the numbering is based on a genotype 1a reference sequence, both the Sequence Locator tool (US database) and the Number tool (European database) are sufficiently flexible to easily accommodate other genotypes. The tools align sequences of all genotypes described to date unambiguously to the reference sequence, so the numbering will be uniform for other genotypes as well as for genotype 1. This numbering proposal, using the AF009606 (isolate H77) sequence as a reference, should be able to unequivocally number all possible mutations in HCV, both natural and manmade. The HCV sequence databases8 and the Los Alamos HCV immunology database9 (as well as the Los Alamos HIV database) number positions and epitopes according to this system. Moreover, the databases websites provides tools for finding stretches of sequence by their numbers, for assigning start and end coordinates to a sequence, and for converting between the various numbering systems. Numbering HCV nucleotide sequences is done by analogy to H77. The first step is aligning your sequence to H77. If there is no length variation, the numbering is straightforward; nucleotide numbers run from 1 (start of 5′ UTR) to 9646 (end of 3′ UTR). Insertions relative to H77 are labeled with letters. Protein numbering works like the nucleotide numbering, but starts at the start of the polyprotein. The sequence databases will support both systems, but use polyprotein numbering as a basis. Absolute numbering moves across the coding regions, relative numbering starts over at every coding region. Relative numbering is almost exclusively used for proteins, polyprotein numbering mainly in immunology, protein numbering in drug resistance research. The Los Alamos immunology database uses polyprotein numbering. The 5′ UTR numbering starts at 1 and ends at 341; the Core cds starts at 342. The numbering of the 3′ UTR starts at 9378 (after the stop codon), but complications arise due to the variable length of the PPT. The UTR consists of 3 elements: a variable 5′ region, the PPT, and a conserved 3′ region, often called X. The first region is numbered 9378–9410. The PPT consists almost entirely of T's and therefore cannot be meaningfully aligned; it is numbered according to its length in H77, 9411–9545. The X region starts at 9546 (regardless of its actual location, which depends on the length of the PPT) and ends at 9646. The HCV databases can be found at: European site: http://euhcvdb.ibcp.fr Japanese site: http://s2as02.genes.nig.ac.jp/ American site: http://hcv.lanl.gov The European HCV database is funded by HepCVax Cluster (EU FP5 grant QLK2-2002-01329) and VIRGIL Network of Excellence (EU FP6 grant LSHM-CT-2004-503359). The Japanese Hepatitis database is supported by a Grant-in Aid for the Publication of Scientific Research Results (#168111), Grant-in Aid for Scientific Research, Japan Society for the Promotion of Science (JSPS). The American HCV database is funded by the Division of Microbiology and Infectious Diseases, National Institute of Allergy and Infectious Diseases under interagency agreement Y1-A1-1500.