Nuclear factor of activated T-cells, NFAT

Discussion

Introduction
Gerald Crabtree
H. Moo Kwon
Anjana Rao
Stephen Harrison
Tom Kerppola
Cornelis Verweij
Stuart L. Schreiber
Steffan N. Ho
Neil Clipstone
W. Michael Flanagan
Naoko Arai
PJ Utz

Introduction

The proposal to change the  nomenclature scheme for the genes encoding members of the NFAT family of transcription factors has provoked a number of responses and comments. The history of identification and naming of the various genes is clearly open to different interpretations, and the nomenclature committee has decided that it is counterproductive to publish extensive accounts of the history as seen by the key participants. Whatever has happened previously what is now at issue, from our perspective, is the clear and unambiguous naming of the genes concerned. I have therefore tried to summarise below the main scientific points made in the discussion. I hope that the NFAT research community will be able to reach some consensus about the names to be used, to prevent further confusion in the literature and databases. However, if the community is unable to agree, the nomenclature committee must at some point reach a decision over the approved gene symbols. Please remember these are only about 5 of the estimated 80,000 genes that we have to deal with!
Julia White

Summary (more comments will be added when agreed with submitters)

Please note Disclaimer.

Gerald Crabtree
The scientific point at issue is whether to continue the designation of NF-ATc as mechanistically related proteins that are calcium-regulated signaling molecules and transcriptional activators or expand the designation to include more distantly related proteins of the rel family such as TonE-BP, which mediates the cellular response to osmotic stress.  I fear that grouping these proteins, one of which mediates Ca2+ responses, the other mediates osmotic responses will create unnecessary confusion.

In 1995 NF-ATc# was adapted to designate proteins that are mechanistically related in that they are Cytoplasmic and translocate to the nucleus after a Calcium/Calcineurin-stimulus.  Their translocation is Cyclosporin-sensitive.  They have a distinctive series of SPXXSPXXSPXXXD/E repeats (SP-repeats) and also a rel domain.  The SP repeats are necessary for Calcium-dependent cytoplasmic-to-nuclear translocation.  Thus they are both signal transducers and regulators of transcription.

TonE-BP/NF-AT5 (originally reported by Miyakawa and colleagues (AF089824, and J.  Biol.  Chem.  273, 20615-20621, 1998) binds to sequences in genes sensitive to osmotic signaling.  There is no evidence that these genes respond to calcium signaling.  Neither is there evidence that NF-ATc responsive genes respond to osmotic signals. Hence there is no evidence of functional or mechanistic vivo overlap between TonE-BP and NF-ATc family members.  TonE-BP is similar to NF-ATc family members only in the rel domain.  By sequencing the adjacent regions and high resolution mapping we find that TonE-BP/NF-AT5 is not next to NF-ATc3, but rather many genes away.  Hence it has picked up a rel domain by exon recombination and not by duplication and divergence. Thus, TonE-BP is not a homologue by accepted criteria (Doolittle et al). Most importantly, it lacks the SP repeats of other NF-ATc proteins. Finally, TonE-BP is not calcium-regulated, does not translocate from cytoplasm-to-nucleus, is not a calcineurin substrate and is not cyclosporin-sensitive.  Hence it has neither mechanistic nor extensive structural similarity to the other NFATc family members.  To my view a nomenclature that groups TonE-BP with NF-ATc proteins would be like grouping the glucocorticoid receptor with HNF-4, both of which have a ligand binding domain and zinc fingers and hence are more structurally related than TonE-BP and NF-ATc family members.

Naming proteins by their domains leads to an unacceptable terminology because domains are used in combinations and one can not make an unbiased judgement about which domain is most important.  The most distinctive domain and the original feature of the NF-ATc proteins are the SP-repeats that controls their calcium/calcineurin-dependent movement to the nucleus.  The present nomenclature recognizes this fact.  Regarding NF-ATn proteins; we have a paper in press in Nature that shows that the partner protein for NF-ATc4 is not AP-1 and appears to be unknown proteins (almost certainly the case since only about 1/50th of the estimated transcription factors have been characterized.  Also, AP-1 does not account for the tissue-specific expression of transcription directed by the IL-2 NFAT1 site reported by ourselves and Richard Flavell's laboratory.  Thus addition proteins are essential components of the NF-AT1 complex.  The present nomenclature recognizes these find points.

Suggested alternatives:
The "T" we put into the name fails to recognize the role of these proteins is in other cell types. One unbiased solution would be to change the name to reflect their roles in calcium signaling such as:
CALSTATINS-1, 2, 3, and 4  for Calcium Signal Transducers and regulators of Transcription.

H. Moo Kwon
1.  It is clear that TonEBP/NFAT5 is different from NFAT1-4 in many ways except for a low level sequence homology in the rel-like domain.
2.  Our data published paper (PNAS 96, 2538) where we reported the cloning of TonEBP and other work (see below) demonstrate differences between TonEBP and NFATs. (i) The time course of activation of TonEBP in reponse to hypertonicity is very slow - more than 10 hours to complete (ii) The increase in the nuclear TonEBP is due to a combination of an increased abundance of TonEBP and an increased nuclear distribution (iii) Cyclosporin A does not appear to affect TonEBP because expression of downstream genes is not altered.
In summary, I argue that this protein should be named TonEBP (tonicity-responsive enahncer binding protein) rather than NFAT5.  Naming of TonEBP is based on its function.  The only connection to NFATs is a low level sequence homology in the rel domain.

Anjana Rao
NF-AT5/TonE-BP
Despite the fact that it is not cytoplasmic or obviously calcineurin-regulated in the cell types examined so far, NFAT5 is clearly a member of the NFAT family:  it binds an identical site, conserves all the DNA-binding residues, and is evolutionarily closer to NFAT (41-45% identity) than Rel (~17% identity), as also suggested by the fact that its chromosomal location is very close to that of NFAT4. NFAT5, like the other NFAT proteins, will be found to have one or more nuclear partners with which it cooperates strongly at specific composite sites, this parallel with the other NFAT proteins only accentuates the arguments for including NFAT5 in the NFAT family.
On the assertion that NFAT proteins that possess a histidine instead of an arginine in the DNA binding loop will not bind DNA independently: Mutation of this histidine to arginine was certainly required to confer DNA-binding activity on a small (176 aa) subfragment of the NFAT2 DNA-binding domain, but this mutation was originally introduced by Greg Verdine's laboratory for the sole purpose of obtaining an NFAT2-DNA complex of sufficiently small size and stability so that its solution structure could be solved by NMR [4].  When the complete (297 aa) DNA-binding domain of NFAT is examined, its DNA-binding affinity (~20 nM) is neither improved nor diminished by this substitution.

NF-ATn
The "NFAT-1" complex originally described [by Crabtree] formed on a DNA element that has been shown to bind a complex of an NFAT protein and an AP-1 (Fos-Jun or Jun-Jun) dimer. The properties of these dimers match exactly with those defined for NF-ATn. It is unlikely that another, substantially different, NF-ATn exists for this particular site. NFAT proteins can bind other types of sites (e.g.  NFkB/Rel-like sites), and they, like other transcription factors, are likely to have multiple nuclear partners with which they cooperate to form transcriptionally active "enhanceosome" complexes.  Presumably some subset of these nuclear partners will participate in truly cooperative protein-protein interactions with the NFAT proteins on other types of composite DNA elements. Such cooperating nuclear partners [have historically been called] by the actual names by which they are known to the scientific community (e.g.  GATA-4, Fos/Jun).

Stephen Harrison
NF-AT5/TonE-BP
Defining NFAT and NFkB differences by the R to H trade is not, in my view, the right way to look at it.  The R/H swap is an interesting, but rather incidental, difference.  The DNA binding domain of NFAT5 (or whatever it will be called) is an NFAT sibling and only an NFkB cousin, not an equally related cousin of both.  Others who have spent time looking at sequences and their relationship to structure would (I believe) agree.  So the conclusion is: there is a calcium-independent (and calcineurin independent) transcription factor with strong relationship to the 4 NFAT(c)'s. It is not objectively correct to deny the close relationship.  I would therefore argue that the new member deserves an NFAT name; since none of the other molecules is restricted to T cells, and the original meaning of "NFAT" is therefore presumably not relevant.

Naming proteins by domains: On the issue of how to name transcription factors, members of a well-known set of homeodomain proteins are known as Hox1, 2 etc.  So there IS precedent for naming some transcription factors after their DNA-binding element.  Likewise, I suspect, with GATA1,2,3,4. Whether precedent makes something a good thing is another question, of course.

Tom Kerppola
The goal of scientific nomenclature should, in my opinion, be simplicity and utility.  This goal is best served by naming proteins with related structures or functions based on a common root.  There may be disagreement with regard to level of similarity required for inclusion within a group.  However, it seems reasonable to associate each protein with the group that is most similar to it even in cases such as NFAT5/TonE-BP where the level of similarity is lower than among
other proteins in the NFAT group.  The DNA binding domain is generally the most highly conserved and best understood domain of transcription regulatory proteins.  It is therefore the most appropriate region for classification purposes.  It is certainly desirable that nomenclature would also reflect protein function.  However, transcription regulatory proteins generally have different functions in different cell types.  It may therefore not be possible to develop a nomenclature that would reflect all of the functions of any particular protein.

Cornelis L. Verweij
Presently, the NFATc family members are mechanistically related by virtue of  the fact that they are all calcineurin substrates that undergo cytoplasmic to nuclear translocation in response to calcium signals. They are structurally
related by virtue of the SP-repeats and calcineurin binding regions in the N-terminus of the protein.  Classification of these proteins based on these criteria proofed to be logical and practical. Opening the nomenclature to  proteins such as TonE-bp which is osmotically regulated would obscure the mechanistic similarity and cause confusion. Therefore, I feel there is no need to change the present nomenclature.

Stuart L. Schreiber
The calcium-calcineurin-NF-ATc signaling pathway was molecularly revealed in 1991 by the discoveries of (1) calcium-calcineurin as the target of FKBP12-FK506 and cyclophilin-cyclosporin and (2) NF-ATc as the FK506/cyclosporin-sensitive downstream target of calcineurin's actions, resulting in a cytoplasm-to-nucleus translocation.  The field was opened in 1988 by the discovery of an FK506- and cyclosporin-sensitive transcription factor named NF-ATc that binds to the IL-2 enhancer/promoter.  In recent years the calcium-calcineurin-NF-ATc pathway, involving several NF-ATc family members, has been shown to be involved in heart valve development, in cardiac hypertrophy (a major killer of humans), and in signaling in the hippocampus (likely relevant to the acquisition of memory).
The hallmark of each NF-ATc family member is their ability to link the calcium-calcineurin signal to specific NF-ATc-dependent promoters. Including TonE-BP/NF-AT5 in this family therefore makes little scientific sense, and in turn creates unnecessary confusion.  In addition, the long and well-established history of this field and its NF-ATc nomenclature are firmly entrenched in the scientific community.

Steffan N. Ho
The "NFAT" nomenclature issue seems to revolve around utilizing either a function-based nomenclature or a domain / structure-based nomenclature. There are certainly good arguments to be made both for and against either position, not to mention an abundance of examples in which either approach has been used, as well as examples in which a combination of the two approaches has been used. Hopefully there would be no argument that in any system of nomenclature, the primary consideration should be that of facilitating and enhancing communication, and by logical extension, minimizing confusion.

 While the TonEBP/NFAT5 DNA binding domain (DBD) is clearly similar to the DBD of NFATc1-4 (~40% amino acid identity), it is also distinctly dissimilar relative to the high degree of similarity among NFATc1-4 (~80% amino acid identity). Furthermore, current published data, including the structure of the NFATc1 and NFATc2 DBDs, indicate that the differences observed between the TonEBP/NFAT5 DBD and the NFATc1-4 DBD are functionally relevant. Therefore, from both a structural and functional standpoint, it would seem reasonable NOT to use a nomenclature that simply extends the numbering system from 4 to 5, as this would not reflect the unique nature of the TonEBP/NFAT5 DBD relative to NFATc1-4. At the same time, it would also perhaps NOT be reasonable to use a nomenclature that is only based on function (e.g. TonEBP) as this would not reflect the structural similarity to the NFAT DBD as defined by NFATc1-4.

 At the risk of further complicating the matter, perhaps a reasonable compromise might be to refer to TonEBP/NFAT5 as "NFAT-related protein of 220 kD", or "NRp220" ¹. A name such as this would reflect the structural similarity to NFAT proteins, and yet also clearly indicate the unique nature of the protein. Furthermore, assigning new biologic functions to a protein named NRp220 would be less confusing than attempting to assign new biologic functions to a protein named "tonicity enhancer binding protein" or TonEBP. Also, this would allow NFATc1-4 to retain a unique nomenclature which would reflect their unique function as calcium signal transducers and activators of transcription.

Neil Clipstone
It is generally accepted that NFAT proteins share several cardinal functional and structural features that help define them as a unique family of transcription factors.  These features include: (i) their cytoplasmic location, (ii) their exquisite calcium/calcineurin sensitivity, (iii) their inhibition by CsA and FK506, (iv) the presence of the conserved regulatory SP repeat motifs and serine rich domain/region (SRD/SRR),  (v) the conserved Rel homology  DNA binding domain, and (vi) a conserved calcineurin binding site. Hence it seems reasonable to use these well defined criteria to establish membership in the  NFAT family. On that basis, while it is clear that NFAT5 is related to the NFAT family of transcription factors by  virtue of similarities within  their  Rel homology domains, it is apparent that NFAT5 lacks many of the principal features of the NFAT family listed above. Hence I believe that while NFAT5 is clearly related to the NFAT family, it does not appear to satisfy the criteria to qualify as a fully fledged member .
 Lastly, one of the obvious problems in the NFAT field that needs to be addressed, is that there are currently several competing  nomenclatures in use that assign different names to the same NFAT proteins (e.g. NFAT1/NFATc2/NFATp and NFAT2/NFATc1/NFATc). Not surprisingly this has led to substantial confusion in the literature. From a personal perspective I favour maintaining the currently approved NFATc#  nomenclature, since this emphasizes the cytoplasmic location and calcium/calcineurin sensitivity of these factors. However, although I certainly favour keeping the NFAT name, if there is no sign of an easy and mutually agreeable resolution regarding these  different nomenclature conventions, perhaps the best course of action would be to rename the proteins. For what it's worth my suggestions would be: CDAT's: calcium/calcineurin- dependent activators of transcription, or CRAT's:  calcium/calcineurin-regulated  activators of transcription.

W. Michael Flanagan
I favor the original nomenclature for the following reasons.
The activity initially described as NF-ATc has been shown to be encoded by at least four genes given the offical names of NF-ATc1-4. These genes are mechanistically related by the fact that all of them are initially Cytoplasmic, translocate after a Ca2+ stimulus, are Calcineurin-dependent and Cyclosporin-senstive. They are structurally related by virtue of the fact that they have a calcineurin binding site, have a series of repeats of a novel motif, SPXXSPXXSPXXXXXD/ED/E and a DNA binding domain similar to a rel domain. Calcineurin dephosphorylates the serines in the SP repeats of each family member. Thus, they are mechanistically and structurally related and should be named in accord with these mechanistic and structural similarities. For this reason I favor the continuation of the originally nomenclature, NF-ATc1,c2 , c3, c4 etc. New members should be defined by the following:

Structural features
1). SPXXSPXXSPXXXXXD/E repeats involved in cytoplasmic -to-nuclear translocation.
2) calcineurin binding site
3) rel domain (however other DNA binding domains could also be acceptable, since yeast use a NF-AT-like protein that has a zinc finger)

Mechanistic features:
1) Translocate from the cytoplasm to the nucleus
2). Translocation is Ca2+/Calcineurin-dependent
3) Translocation is blocked by CsA and FK506
4) They are calcineurin substrates

Naoko Arai
 My thoughts about the nomenclature of NFAT-like proteins is tempered by the thought that it may still be premature to decide on permanent names that functionally distinguish these factors.  We all are certainly aware that the family of NFAT-like factors have overlapping (redundant) or distinct functions under different physiological conditions. For example, under the simple assay of IL-2 reporter gene transactivation, NFAT family members behave indistinguishably. However, on a logical footing, it's only reasonable that the various NFAT-like genes are consistently named in order of their discovery--as has been suggested, with some conflict between who's 1 or 2!--for record-keeping, without losing their "comfortable" aliases (i.e., NFATx, c, p, etc).  Certainly the mechanisms that track genes and sequences allow for this solution.  Sequence variants--which are numerous!--due to alternatively-spliced NFATs can be dealt with as other genes, purely at a gene sequence level.   The most recent addition to the family is the NFAT5 ( human gene isolated by Rao group; we have also cloned this gene and call it NFATz; manuscript in preparation) that is undeniably a member of the family--albeit distantly--at the structural level.  Alignment of NFAT5 with NFATs1-4 (including structurally-aligned NF-kB sequences) easily shows that NFAT5 groups with the NFAT branch, and indeed serves as its' local "root", or member-most-closely-similar to the putatative NFAT ancestor (following divergence from NF-kB's).

PJ Utz
There are three critical issues in the discussion:
(1) Should NFAT nomenclature be based on structural or functional grounds?
(2) Does TonE-BP meet these criteria?
(2) How will changes affect the ability of future researchers to understand and interpret the literature that surrounds these expanding families of complexes.

Regarding (1), I believe that the NF-AT complex is best named based on its function as a transcription complex that translocates from the cytoplasm to the nucleus following calcium/calcineurin stimuli, in a cyclosporine-sensitive manner.  The terms NF-ATc1,c2,c3,and c4 accurately designate the Cytoplasmic, Cyclosporin-sensitive, and Calcium-dependent
components of these complexes (Flanagan et al, 1991, Nature, 96:2538-42. 352:803). The present nomenclature also recognizes that the critical event underlying biologic specificity of this complex is its combinatorial assembly.  The difficulty with the proposed change in nomenclature is that it fails to recognize the fact that NF-ATc proteins never bind alone, and that NF-AT1 is a transcription complex, not a subunit. In fact, review of the abstract of our original description specifically identified NFAT as "a protein complex, NFAT-1."  Therefore I remain a strong advocate for the continued use of the current nomenclature that is based on the functional characteristics of the NF-AT transcription complexes.

Regarding (2), by virtue of the presence of SP repeats, the current NF-ATc# nomenclature would by inference incorporate the new "domain structure" that has been proposed.  TonE-BP is not calcium/calcineurin inducible, is not sensitive to cyclosporine, lacks an SP domain, and has properties (such as a very slow activation time-course and tonicity-sensitivity) that identify it as a unique complex. This view has been embraced by the discoverer's of this complex (Miyakawa et al., 1999, PNAS, 96:2538-42).  Therefore, it should be named differently than members of the NF-AT family.

Regarding (3), the current nomenclature is easy to understand and is firmly entrenched in the scientific literature.  The present nomenclature also leaves room for the discovery of new partner proteins (NF-ATn) for NF-ATc family members. Finally, it is likely that other complexes with properties similar to TonE-BP will be identified, and the present system leaves room for the addition of these new members as well.

Footnotes (from the Human Gene Nomenclature Committee)
1. The convention for naming genes by their similarity to others is to use "L" for like, ie in this case NFATL1. Molecular weights are usually avoided in gene symbols as estimates are often variable, and hence including it can be confusing. Simple arbitrary numbering is used instead. HGNC Guidelines .

Return to NFAT page