Mitochondrial DNA Sequencing on HV1/HV2 and the rCRS Reference

The control region accumulates mutations faster than the coding portions of the mitochondrial genome because it does not encode a functional product and therefore faces less purifying selection. Within the control region, HV1 and HV2 are designated hypervariable because population surveys have found that a disproportionate share of sequence differences between individuals cluster in these two segments. A comparison of the two reference standards, the rCRS and the RSRS, reveals that HV1 and HV2 together carry the majority of the sequence differences between the Cambridge sequence and the reconstructed ancestral human sequence.

Forensic sequencing of HV1 and HV2 proceeds by PCR amplification of the two regions followed by Sanger-method dideoxy chain-termination sequencing. The amplicons are designed to overlap slightly to confirm that the HV1 and HV2 sequences are from the same molecule. Each sequencing read is compared nucleotide-by-nucleotide to the rCRS, and differences are noted as substitutions, insertions, or deletions relative to the reference position numbers. A result might read: 16126C, 16294T, 73G, 263G, 315.1C, meaning that at position 16126 the sample has cytosine where the rCRS has thymine, and so on. This notation system, standardised by the ISFG (International Society for Forensic Genetics) and the ENFSI mtDNA Working Group, is what allows profiles from different laboratories and different countries to be compared in the EMPOP database.

EMPOP (the EMPOP mtDNA Population Database), maintained at the Institute of Legal Medicine at Innsbruck, Austria, is the forensic community's central reference population resource. As of 2024 it contains over 50,000 sequences from populations worldwide, curated to remove artefactual sequences and phylogenetically classified. In the US, the FBI's mtDNA Population Database (a predecessor to EMPOP integration) has been used in casework since the late 1990s. In India, CFSL laboratories performing mtDNA analysis use reference frequency estimates from published Indian population datasets, with the ENFSI mtDNA Working Group guidelines providing the methodological framework. Across the European Union, ENFSI member laboratories contribute sequences to EMPOP under a data-sharing agreement that ensures consistent quality and nomenclature.

A third hypervariable region, HV3 (approximately positions 438 to 576), is sometimes sequenced in cases where the HV1/HV2 profile is insufficiently discriminating, but this is not standard in routine casework. Whole-mitochondrial-genome sequencing, now available via MPS (massively parallel sequencing), provides the highest resolution: instead of 610 base pairs of HV1/HV2, the analyst sequences all 16,569 positions, dramatically improving discrimination. The Verogen ForenSeq mtDNA Whole Genome Solution is the commercial kit used for this purpose in high-resolution casework in the US, UK, and EU.

The original Cambridge Reference Sequence (CRS) was published by Anderson et al. in Nature in 1981 and was the first complete human mitochondrial genome sequence. It was derived from HeLa cells and from a placental sample of European origin. Because it was published first, the entire forensic community adopted it as the comparator for position numbering, and that number scheme is embedded in decades of published casework, population databases, and laboratory protocols.

Two problems emerged. First, the 1981 sequence contained several errors that were identified when the material was re-sequenced. In 1999, Andrews et al. published the corrected sequence, designated the revised Cambridge Reference Sequence (rCRS), deposited in GenBank as NC_012920. The rCRS is now the mandatory reference in EMPOP and in ISFG guidelines, and all modern forensic reports express differences relative to the rCRS. An analyst using the original 1981 CRS rather than the rCRS will produce a notation that differs at 11 positions, which could cause a false apparent discrepancy in a cross-laboratory comparison.

Second, the rCRS is not a universal human ancestor: it is a modern European sequence, and variants common in Asian or African populations appear as substitutions relative to the rCRS even though they represent ancestral states. Behar et al. (2012) proposed the Reconstructed Sapiens Reference Sequence (RSRS), derived from phylogenetic analysis to represent the most probable ancestral sequence of all living humans. The RSRS is used in academic phylogenetic research and in some population-genetics analyses, but it has not displaced the rCRS in forensic practice because the entire legacy database and nomenclature system is built around rCRS coordinates. ISFG guidelines maintain rCRS as the forensic standard.

In operational casework the implications are practical. A UK laboratory using EMPOP must report results as differences from the rCRS. A US FBI laboratory must use the same convention. An Indian CFSL laboratory reporting under NABL accreditation cross-referencing ENFSI guidelines uses the rCRS. A result from a Vietnamese identification laboratory submitted to DPAA (Defense POW/MIA Accounting Agency) for family comparison must be in rCRS notation to be matched against the AFDIL reference database. Any laboratory that deviates from rCRS notation must include explicit conversion information in its report, or the cross-laboratory comparison is compromised.

Heteroplasmy refers to the coexistence of two or more different mtDNA sequences within a single individual. It arises because somatic mutations in the mitochondrial genome can become clonally expanded in a tissue. Heteroplasmy can be present at low frequency (a minor variant that appears as a shoulder on a Sanger sequencing peak), at high frequency (two peaks of comparable height), or anywhere in between. It can be tissue-specific: an individual may be heteroplasmic in hair follicles but homoplasmic in blood, because the two tissues had different lineage histories during development.

The forensic consequence is significant. If a questioned hair and a blood reference sample from the same individual are compared, and the hair carries a heteroplasmic position that the blood sample does not, a naive analyst might call an exclusion. The correct interpretation is that the hair evidence and reference are consistent: the heteroplasmic variant in the hair represents a somatic mutation that has not been clonally expanded to detectable levels in blood. The SWGDAM Interpretation Guidelines for Mitochondrial DNA (2019 revision) and the ENFSI mtDNA Working Group Best Practice Manual both address this: an apparent single-nucleotide difference between a questioned sample and a reference can be attributed to heteroplasmy if the variant has been observed in the relevant tissue type in the published literature, and the analyst must document the reasoning.

Point heteroplasmy (a single position where two nucleotide variants are observed) is more common and easier to interpret than length heteroplasmy, which occurs in the C-stretch at positions approximately 303-315 and 16184-16193. These cytosine-stretch regions are prone to replication slippage, generating length variants (extra or missing cytosine residues) that produce a complex overlapping sequencing pattern. Length heteroplasmy in these regions can be documented as a common variant rather than an artefact, but it requires particular care in Sanger sequencing and is one reason some laboratories prefer MPS for its ability to resolve length variants as distinct reads.

In the identification of Tsar Nicholas II and the Romanov family (1998 confirmation, following the initial 1994 identification work led by Dr Peter Gill and the UK Forensic Science Service), a critical challenge was exactly this: Tsar Nicholas II's mtDNA carried a point heteroplasmy at position 16169 (a mixture of T and C), observed in both his skeletal remains and in a blood sample from his living maternal-line relative Prince Philip, Duke of Edinburgh. The initial 1994 analysis by Dr Gill's team at the Forensic Science Service in Aldermaston treated the result cautiously; by 1998, independent analysis by Russian and US teams confirmed that the heteroplasmy was genuine and consistent with a shared maternal lineage, confirming the identification.

The Romanov identification is the textbook example of forensic mtDNA applied to historical remains under genuine scientific uncertainty. The skeletal remains of nine individuals recovered from a shallow grave near Yekaterinburg, Russia, in 1991 were subjected to STR and mtDNA analysis. Dr Peter Gill and colleagues at the UK Forensic Science Service published the primary identification results in Nature Genetics in 1994. The mtDNA from bones attributed to Alexandra Feodorovna, the Tsarina, and her three daughters matched a reference sequence from Prince Philip, Duke of Edinburgh, a maternal-line descendant of Alexandra's mother Princess Alice of Hesse. The Tsar's bones showed a heteroplasmic position 16169 matching both the skeletal remains and a reference blood sample from Prince Philip. The 1998 confirmation by an independent Russian-US team resolved the last doubts and led to the official state reinterment.

Anna Anderson had claimed since 1920 to be the Grand Duchess Anastasia, the youngest Romanov daughter presumed to have escaped the execution. She died in 1984. In 1994, a post-mortem tissue sample from Anna Anderson's intestinal biopsy, preserved at the University of Virginia Medical Center, was available for testing. Dr Mark Stoneking and colleagues at Pennsylvania State University sequenced her mtDNA and compared it to the reference sequence from Prince Philip and to the Romanov bones. The sequences did not match. Instead, Anna Anderson's mtDNA matched that of a Karl Maucher, a living maternal-line descendant of Franziska Schanzkowska, a Polish factory worker who had disappeared around 1920. The identification confirmed that Anna Anderson was Franziska Schanzkowska, not Anastasia. The case is now a standard reference in discussions of the statistical and interpretive value of mtDNA exclusion in historical claims.

The Mengele identification presented a different challenge: remains recovered in Brazil in 1985 without any living direct relative available as a reference. The initial 1985 examination by a team including Dr Clyde Snow and American forensic anthropologists provided anthropological and odontological evidence consistent with Josef Mengele, but the identification was contested. In 1992, Dr Alec Jeffreys (inventor of DNA fingerprinting) and Brazilian scientists conducted mtDNA sequencing on the 1985 bones and compared the result against a blood reference from Rolf Mengele, Josef's son. A son inherits his mitochondrial genome from his mother, who would have transmitted the same sequence to Josef Mengele as well (assuming the same maternal lineage). The mtDNA sequences matched, providing the molecular confirmation that resolved the case. The 1992 Mengele confirmation became an early demonstration that forensic mtDNA analysis with a single indirect reference sample can provide legally meaningful evidence.

Across the Atlantic, the AFDIL laboratory at Dover Air Force Base has processed over 1,800 mtDNA identifications since the late 1990s as part of the US government's Vietnam War MIA recovery programme. Recovery missions in Vietnam, Laos, and Cambodia retrieve skeletal fragments from crash sites and graves; nuclear STR is routinely unavailable because the remains have been in tropical soil for decades. AFDIL's mtDNA casework protocol, which uses whole-mitochondrial-genome sequencing for higher resolution, is the largest systematic application of forensic mtDNA sequencing in history.

The forensic statistic for an mtDNA match is the profile frequency: the proportion of individuals in a relevant reference population who carry the same sequence (or, more precisely, the proportion who cannot be excluded as the source). This is fundamentally different from the random-match probability in autosomal STR analysis. Because all maternally related individuals share the same haplotype, the mtDNA result cannot individualise; it can only include or exclude. The frequency estimate bounds the inclusion: if the observed haplotype occurs in 1 in 1,000 individuals of the relevant population, the examiner can say only that 1 in 1,000 unrelated individuals could not be excluded, not that the sample came from a specific person.

Population frequency estimates come from EMPOP, the primary curated forensic mtDNA database maintained at the Institute of Legal Medicine, Innsbruck. EMPOP uses a phylogenetic method for frequency estimation rather than simple counting: sequences are placed on the human mtDNA phylogenetic tree, and frequency estimates are derived from the population distribution of the haplogroup and sub-haplogroup to which the profile belongs. This approach reduces the sensitivity to small reference sample sizes and is the standard endorsed by ISFG and ENFSI.

In the United States, the FBI's population database preceded EMPOP and has been used in federal and state courts under Daubert and state equivalents. Expert testimony on mtDNA frequency has been challenged in multiple US cases; the most significant was United States v. Beverly (2003) and a series of state-level Daubert hearings in the early 2000s. In the UK, ENFSI member laboratories contribute to EMPOP and report frequency estimates from the European dataset stratified by haplogroup. In India, CFSL Hyderabad and CFSL New Delhi perform mtDNA analysis, and published Indian population datasets provide the reference frequencies, though the absence of a national mtDNA database analogous to NDNAD or EMPOP remains a gap in the quality framework.

The minimum reporting standard, required by SWGDAM (US), ENFSI (EU), and the Forensic Science Regulator (UK), is that any mtDNA report that results in an inclusion must include a frequency estimate from a relevant population database, the number of sequences in the reference database, and the method of frequency calculation. A qualitative inclusion statement without a frequency number does not meet any of these standards. In Indian NABL-accredited labs, the reporting requirement follows ISO/IEC 17025 section 7.8, which mandates that results be reported with their associated uncertainty, which in the mtDNA context is the frequency estimate and its database provenance.