Letter to the Editor
Nature Genetics
545 National Press Building
529 14th St. N.W.
Washington, D.C. 20045

May 15, 1997

Sir - The article by Fleury et al. in the March issue of Nature Genetics(1) on UCP2 
received much media attention due to its potential as a target for obesity controlling drugs.  
This may be very well deserved, however, the uncoupling protein UCP1 and UCP2 are not 
the only genes in humans that must be considered.  An extensive search of the Genbank 
EST section and the TIGR database uncovered two more closely related genes.  Therefore, 
UCP1 and UCP2 are only half the picture.  One of these (UCP3) is very similar to UCP2 
and the other (UCP4) is most like UCP1.  

The UCP1 and UCP2 genes are are 59% identical at the amino acid level.  They can be 
found in Genbank (U28480 and U82819).

UCP3 (TIGR database THC182634 = Genbank AA192136 and Z28895) is similar to 
UCP2, but it is a different gene.  It is 70.4% identical to UCP2 over the last 115 amino 
acids, with its cDNA found in human muscle.  The translated C-terminal fragment starts at 
the end of transmembrane segment 4.  This sequence differs from both mouse and human 
UCP2, so it is not due to a contamination problem.

UCP4  (TIGR database THC164771 = Genbank AA065221 and AA064919) is an isolog 
of UCP1 found in human fibroblasts. The translated C-terminal fragment begins at the end 
of transmembrane segment 3.  It is 83.7% identical to UCP1 over 178 amino acids.  
Comparison to mouse shows this sequence is identical to the mouse UCP sequence at the 
amino acid level (M21222, M21244 to M21247).  This raises the question of possible 
contamination of a human library with mouse DNA.  Examination of the DNA sequences in 
exon 5 and 3' of the coding region in exon 5 shows that the DNA sequence is identical 
even in the 3' untranslated region.  Consequently, there must be a contamination problem.  
Since both Genbank entries of the human gene are 5' and 3' reads of the same clone 
zm51f07, it is probable that the Stratagene fibroblast library #937212 is contaminated with 
mouse DNA and UCP4 is really just mouse UCP1.

The excitement over UCP2 needs to be spread out to include UCP3, which may be equally 
important.  Futhermore, the great expansion of EST databases requires mechanisms to 
identify and remove, or more properly, correctly label contaminants like the UCP4 
sequence.


David R. Nelson
email dnelson@utmem1.utmem.edu
FAX (901) 448-7360


References

1.  Fleury et al. Nature Genetics 15, 269-272 (1997)


Addendum July 8, 1997

Nature Genetics declined publication of this letter because the sequence of UCP3 has been 
published in two papers by Boss et al., FEBS 408, 39-42 (May, 1997) and Vidal-Puig 
et al., Biochem. Biophys. Res. Commun. 235, 79-82 (June, 1997).  This makes the above 
observations unpublishable, so I include it here on my own web site.

Another sequence from the Stratagene human fibroblast library (#937212) is actually 
mouse.This is the EST AA100010 with the best match to the mouse Cyp4a10 sequence 
Pvalue = 2.3 e-96 at the protein level and 1.0 e -160 at the nucleotide level.