Replication of linear adenovirus DNA is not hairpin-primed

Abstract

THE mechanism of replication of the 5′ ends of linear DNA chromosomes remains unclear^1,2, as the known DNA polymerases, which act in the 5′ to 3′ direction, require a free 3′-OH as a primer for chain elongation. If a short ribonucleotide sequence primed DNA synthesis, its subsequent removal would leave no primer for gap fill synthesis at the 5′ end. Most linear bacteriophage DNA molecules therefore have terminal repeats or single stranded complementary ends, which allow formation of circular or concatemeric replication intermediates¹. Several interesting models have been proposed to explain replication of the 5′ ends of eukaryote chromosomes and linear DNA molecules which cannot circularise. Cavalier-Smith³ proposed that chromosome ends can form self-complementary hairpin loops, allowing the parental 3′ end to act as a primer for DNA polymerase, followed by endonuclease action and repair synthesis. This model has been modified by Bateman⁴ and Tattersall and Ward⁵ have extended it to propose a rolling hairpin model for the replication of parvovirus and linear chromosomal DNA. There is now considerable evidence for hairpin priming of replication of the DNA of minute virus of mice (MVM) and adeno-associated virus (AAV)^5–7, and hairpin priming may well be a general mechanism for replication of linear DNA molecules that are unable to circularise or form concatemers^3,5,7. Roberts⁸ and Wu et al.⁹ have proposed a hairpin-primed model for replication of adenovirus DNA, which is summarised in Fig. 1. We present evidence here which is not consistent with a hairpin-primed model for replication of adenovirus DNA.