Every organism’s DNA has a phosphate backbone, which allows the DNA molecule to fold into itself. This keeps hydrophilic phosphates on the outside of the molecule, and protects the hydrophobic nucleotides which code for proteins. Phosphorous is a key element necessary for life.
In December 2010, NASA scientists announced the discovery of bacteria in California’s Mono Lake (pictured above) that incorporate arsenic into its DNA instead of phosphorous. Remember the big fuss about that story? (If not, see David Klinghoffer’s post here.)
The announcement was over-hyped and heavily criticized. The hype centered on implications that arsenate-containing DNA were thought to have for some of NASA’s top unanswered questions. These findings, we were told,

1. May have implications for the search for extraterrestrial life;
2. May have implications for origin of life scenarios, including the origin of life on other planets;
3. May dispel notions that there are certain compounds that are necessary for life.

But any chemist worth his salt knows that while arsenic and phosphorous are within the same group on the periodic table, that does NOT mean their properties are identical. In the case of phosphorous and arsenic, their bonding structures can be similar, but phosphorous is generally more stable than arsenic, meaning phosphate (PO₄^-3) is more stable than arsenate (AsO₄^-3). Furthermore, arsenic is substantially larger (on a molecular scale) than phosphorous, which would stress the three-dimensional structure of the DNA molecule as well as any associated bonding with other proteins.
Even if bacteria had incorporated arsenate into its DNA, something that was not definitively proved when the announcement was made, there is no reason to believe that this structure is as stable as its phosphate alternative, and certainly no reason to believe that it leads to protein translation.
Several scientists voiced their concern over this and other issues with the study. Casey Luskin documented them here.
Now, almost two years later, Nature confirms what most of us already figured was true: “‘Arsenic-life’ bacterium prefers phosphorous after all.” This paper addresses one of the unconfirmed questions regarding this study: How do the bacteria know the difference between phosphate and arsenate, and do the bacteria prefer one over the other? Those questions have been laid to rest. Not only do the bacteria preferentially consume phosphate, but they are so desperate for the stuff that they will fish out the smallest bit of phosphate in a sea of arsenate. From the report:

Their threshold for when ‘discrimination’ broke down was when 50% of the proteins ended up bound to arsenate — indicating that the ability to discriminate had been overwhelmed. Even in solutions containing 500-fold more arsenate than phosphate, all five proteins were still able to preferentially bind phosphate. And one protein, from the Mono Lake bacterium, could do so at arsenate excesses of up to 4,500-fold over phosphate. (emphasis added)

The only reason the bacteria incorporated arsenic at all was because there was nothing else around; as soon as it found phosphorous, it readily replaced the arsenic.
So it seems the hype was indeed just that, hype. As one critic comments:

“The latest paper shows that the ‘arsenic monster’ GFAJ-1 goes to a huge amount of effort, “even more than other life,” to avoid arsenate, says Wolfgang Nitschke from the Mediterranean Institute of Microbiology in Marseilles, France, who co-authored a commentary questioning the conclusion that GFAJ-1 could replace phosphate with arsenate [reference removed]. “This shows clearly that life doesn’t like arsenate in cytoplasm…”

So, for now, we are still left with life’s odd little proclivity for very specific molecules, and only those molecules, which must be arranged in a particular way.