This has always been a mystery to me. When DNA is replicated, the enzyme DNA polymerase attaches to both of the antiparallel strands and both act as an individual template. One strand is oriented in the 5′ to 3′ direction and the other is in the 3′ to 5′. The latter strand is replicated in short sections of DNA, called Okazaki fragments, whereby the enzyme can run in the 5′ to 3′ direction, consequently forming short sections of DNA that would need to be joined together to form a continuous new strand. This strand, which is made discontinuously, is dubbed the ‘lagging strand’. Synthesis of this strand takes slightly longer than the other strand, the ‘leading strand’.
This method of DNA replication is complicated, and one may wonder why the DNA polymerase doesn’t just work in both directions. One reason for this proofreading.
We have around 25,000 genes in our genome. Each gene can contain thousands of nucleotides (either A, T, C, G) , so when the DNA is being replicated, there is a chance that at any point the wrong nucleotide is added. Despite this huge number of potential errors, the actual error rate is only one in every 10^7 nucleotides.
Why is it so low?
The low error rate is on account of two reasons. One is the complementary base – pairing rules, whereby Adenine forms the strongest bonds with Thymine, and the same goes for Cytosine and Guanine. However incorrect base pairs can also be formed, eg A-C.
How is this prevented?
During replication, DNA polymerase carefully monitors the base pairing. Only when the correct pair has formed will the enzyme continue to replicate DNA. When a mistake is made, it is usually corrected though the aforementioned proofreading.
What is it?
While DNA is being synthesised, a nuclease within the DNA polymerase is ready for use. If an incorrect nucleotide is added, this nuclease clips off the mispaired nucleotide by cleaving the phosphodiester backbone which allows the polymerase to continue adding correct nucleotides to the ever growing chain.
Why is this relevant to the question?
Now, if the DNA polymerase were to hypothetically polymerise DNA in the 3′ to 5′ direction, it would be unable to proofread. Once the nuclease removed an incorrectly bound nucleotide, a chemical dead end would be created, and the chain could no longer be elongated. Thus, for proofreading to occur the polymerisation must occur in the 5′ to 3′ direction
Resources: “Essential cell biology”