6.2. A phylogenetic network for the Finnish mtDNA haplogroup UK

Haplogroup U is determined by a transition 12308A>G in the tRNA^Leu(CUN) gene. Haplogroup U is almost specific to Europeans, and it is found only at a low frequency in the Japanese, the North African Berber population, the Ethiopians and the Senegalese (Ozawa 1995, Torroni et al. 1996a, Passarino et al. 1998, Macaulay et al. 1999). Haplogroup U is composed of subclusters termed U1-U6 and U*, in addition to which haplogroup K has been included as a subgroup (Richards et al. 1998). Estimated minimum divergence times have suggested that haplogroup U is ancient, and the oldest subgroup U5 dates back to about 52,000 YBP (Richards et al. 1998). The divergence times of other European haplogroups vary a lot: 12,500–18,500 YBP for haplogroups V, K and W and about 20,500–28,000 YBP for haplogroups H, X and J. Haplogroups I and T are among the oldest, with divergence times of 35,000 and 46,500 YBP, respectively (Richards et al. 1998).

The phylogenetic network of haplogroup UK (see Figure 4) was constructed in two parts. The first part was constructed using sequence data on 22 population samples from the provinces of Northern Ostrobothnia and Kainuu. This turned out to be exclusively a U5 network (Figure 2a, III). An analysis of 21 control samples with 12308A>G but not belonging to haplogroup U5 was used to create a complete network for haplogroup UK (Figure 1, IV). Comparison of this haplogroup UK network with four previously published complete mtDNA sequences containing 12308A>G (Ozawa 1995; Arnason et al. 1996; Ohlenbusch et al. 1998) revealed that all four sequences could be unambiguously placed in the network suggesting that we have detected all the important polymorphisms that characterize haplogroup UK. Our network turned out to be almost a perfect tree, because only three parallelisms could be identified in the coding sequence. Two of them, transitions 1811A>G and 3197T>C, existed between U5 and other subgroups. Several parallelisms were found in the sequence of HVS-I, however.

Comparison of the expanded network with that based on a compilation of European samples revealed subcluster U2 by identification of polymorphisms at nts 15907, 16051, 16129, 16189 and 16362 (Macaulay et al. 1999). Similarly, subcluster U4 was identified on the basis of substitutions at nts 4646, 11332, 16356 and 16362. Cluster K has been defined by polymorphisms at nts 16224 and 16311 (Macaulay et al. 1999), but we found seven additional common polymorphisms determining this cluster. Substitutions 10398A>G and 11299T>C have been thought to create separate branches (Macaulay et al. 1999), but we found exclusive co-occurrence of these two polymorphisms among Finns. The previous networks (Richards et al. 1998, Macaulay et al. 1999) have also included a group of poorly defined haplotypes designated as cluster U*, which was the most probable designation here for five controls and two patients. The HVS I motifs determining subclusters U1, U3 and U6 (Macaulay et al. 1999) were not found among the Finns examined.

The haplogroup UK network enabled the relationships between the various clusters to be identified. Contrary to CRS, clusters U and K had six coding region variants in common, and thereafter, a reticulation composed of the transitions 1811A>G and 3197T>C divided the haplogroup UK into three groups. 3197T>C was found in subcluster U5, 1811A>G was found in K, U4 and U*, and both transitions were found in subcluster U2. Cluster K was fairly distant from subcluster U5, as the most recent common node was separated from its central node by nine coding region substitutions. A striking feature in the genomes belonging to K, U2, U4 and U* was the considerable number of polymorphisms and the paucity of branching. This may be due to the possibility that branching haplotypes are extinct or to the rarity of these haplotypes in the population.

The character of subcluster U* was then considered further. The Hae III restriction site at position 16517 is a variable one, and a restriction site gain suggesting the presence of 16519T>C has been found in samples belonging to clusters U2, U3, U4, U* and K, whereas it is absent in U1 and U5 and no data have been reported for U6 (Macaulay et al. 1999). We found that 16519T>C had arisen together with 1811A>G and was thus present in all the samples in the network with the exception of cluster U5. Furthermore, a back mutation had taken place in the cluster including patient #411, three controls and the previously reported patient P-8 (Ozawa 1995). Using this additional information, we suggest that there is an additional star-like cluster in the UK network that is defined by the presence of 1811A>G, 16519T>C and the wild type nt A at position 9698 and includes patients #148 and #260 and four population controls. As this cluster includes patient #260, who belongs to subcluster U4, we would like to term this subcluster U4. Furthermore, the subcluster defined by 1811A>G, 9698A>G and the wild type nt T at position 16519 and includes patient #411 and three controls may be proposed as U7. This subcluster is quite interesting in that it is found in populations as diverse as the Finns and the Japanese (Ozawa 1995) and, on the basis of the HVS I polymorphisms at nts 16146 and 16342, in the Basques as well (Richards et al. 1998).

We found that 83% of the controls with 12308A>G belonged to subgroup U5 (Richards et al. 1998, Macaulay et al. 1999), suggesting restricted variation in haplogroup U in northern Finland. The frequency of U5 was found to be 5 to 6-fold higher than in the Germans (Hofmann et al. 1997). The upper branch of U5 is characterized by 5656A>G, and the HVS-I sequence suggests that it is similar to the previously defined subcluster U5b (Richards et al. 1998), which may be quite specific to the Finns, as 5656A>G is detected among Finns approximately 34 times more frequently than in the Hungarians or the English (Thomas et al. 1996, Zsurka et al. 1997, II). Interestingly, the transition 5656A>G was first associated with maternally inherited progressive tubulointerstitial nephritis in a Hungarian family and was not detected among the Hungarian controls (Zsurka et al. 1997). These differences in frequencies suggest that the Finns and the Hungarians are genetically distant in spite of the fact that the Hungarian language belongs to the Finno-Ugric group in the same way as Finnish and Saami languages (Korhonen 1991). Furthermore, part of the U5b branch is characterized by the HVS-I motif, including 16144T>C, 16189T>C and 16270C>T, which has been considered to be fairly specific to the Saami and is not found among other Europeans except for the Finns (Sajantila et al. 1995, Lahermo et al. 1996). The branch characterized by nt 13637A>G appears to be fairly rare in Finns, whereas its most probable counterpart, subcluster U5a (Richards et al. 1998), is not uncommon among other Europeans. These findings suggest that the haplotype characterized by the coding region variants 5656A>G and 12618G>A may be highly specific to the Finns and the Saami.

Geographical, linguistic, and cultural factors have contributed to the isolation of the Finnish population (de la Chapelle 1993, Peltonen et al. 1995) and this may partly explain the limited variation within haplogroup U. It has been shown that the frequency of a mtDNA genotype is higher when it has been introduced into the population earlier (Heyer 1995). Moreover, the rate of mtDNA loss in an expanding population may be as high as 72% within 150 years or in six to eight generations (Heyer 1995), which may explain the different haplogroup frequencies between populations. Moreover, the current frequencies of haplogroups in a population do not necessarily reflect the frequencies a few generations ago, as shown by the example of haplogroup V in the Basques. This haplogroup is observed nowadays only in northwestern Europe and in North Africa. It reaches high frequencies in some Iberian populations and is also very common among the Berbers of North Africa, but showed its highest frequencies (about 41%) among the Saami (Torroni et al. 1998). This observation has been thought to indicate that haplogroup V originated in Europe or North Africa, most likely in the Iberian peninsula, and that it reflects a major late Palaeolithic population expansion from southwestern Europe towards the northeast (Torroni et al. 1998). Dental samples from prehistoric Basques nevertheless show an absence of haplogroup V (Izagirre & de la Rua 1999). Therefore, the date of origin of haplogroup V might be more recent than that proposed (Torroni et al. 1998), or else the observation reflects the high rate of mtDNA loss in the expanding population (Heyer 1995).