PAR bZIP (HLF/NFIL3/DBP)
The PAR bZIP group contains three proline- and acidic-amino-acid-rich basic region-leucine zipper activators - DBP (albumin D-site binding protein), HLF (hepatic leukemia factor) and TEF - together with NFIL3 (E4BP4), which lacks the PAR domain, represses rather than activates, and yet reads the same site. That site is the palindrome TTACGTAA, written in much of the literature as TTATG(T/C)AA. It differs from the cyclic-AMP response element at only the outermost bases of each half-site, TTAC in place of TGAC. DBP, HLF and TEF accumulate with circadian rhythms in liver and suprachiasmatic nucleus, and NFIL3 oscillates in antiphase, so the same element can be activated or repressed depending on time of day. This is the most internally consistent bZIP family in the collection, and its four members are indistinguishable by sequence.
The factors
PAR stands for proline- and acidic-amino-acid-rich, the character of the region N-terminal to the basic region in DBP, HLF and TEF (also called VBP). All three are bZIP activators. NFIL3, better known as E4BP4, is grouped with them because its DNA-binding specificity is the same, but it is structurally distinct: it has no PAR domain and it carries a repression rather than an activation function 11316793,12386933, so the two halves of this family do opposite things at identical sites. Expression is informative. DBP, HLF and TEF accumulate with high-amplitude circadian rhythms in tissues where clock gene expression cycles strongly, notably liver and the suprachiasmatic nucleus, while remaining near-constant in most other brain regions 15175240. NFIL3 cycles in antiphase to DBP, which makes the pair a repressor-activator switch running on one element 11316793. NFIL3 is broadly expressed and prominent in immune cells; HLF is largely restricted, and in human hematopoiesis its expression marks the stem cell state 1386162,34499717. Dimerization matters for interpretation: the bZIP class heterodimerizes extensively, and systematic interaction surveys show these networks are highly partner-specific and have been rewired repeatedly over evolution 23661758, so which sites a PAR protein occupies depends on the partner it has.
The motif
The palindrome is TTACGTAA: two TTAC half-sites in opposed orientation, one per subunit, usually with A, T or G immediately outside, giving GTTACGTAAT and ATTACGTAAC in the highest-information models here 8065331. Much of the older literature writes the site as TTATG(T/C)AA; that form is the same element with one half-site degenerate, and both appear in this collection - HOCOMOCO and JASPAR HLF and NFIL3 models read TTATGTAAC and TTATGCAAC, while the DBP and TEF models read TTACGTAA. The distinction from the cyclic-AMP response element is two bases and no more: TTAC against TGAC at the outer edge of each half-site, with the ACGT core shared. That difference is real but slim, and truncated matrices lose it. This family is unusually coherent. Across 15 models from four databases its internal cohesion is 0.94, the highest of the bZIP families here, and every JASPAR and CIS-BP model in the family sits above 0.97 similarity to another member. Four of the six HOCOMOCO models carry in vitro evidence (HT-SELEX or methyl-SELEX), so the claim that these proteins bind this sequence directly rests on more than ChIP. No member is distinguishable, and the family data flags no member as belonging elsewhere and contains no cross-species inferred model. The one model that drifts is CIS-BP M04016 for HLF, GTGACGTAA, which is half a PAR site and half a CRE; the same is true of CIS-BP M04034 for NFIL3, ATGACGTAAT.
Where it acts
The founding site is the D-site of the rat albumin promoter, from which DBP takes its name 2331750, and the family's best-characterized targets are hepatic metabolic genes, including cytochrome P450 enzymes and pyridoxal kinase 8617210,15175240. DBP, HLF and TEF have similar circadian and tissue-specific expression but measurably different promoter preferences 8617210, so a shared matrix does not imply a shared target list. Because occupancy is rhythmic and NFIL3 occupies the same elements at the opposite phase 11316793, a site can be an activator target in one part of the day and a repressor target twelve hours later; a static motif scan cannot express this. In immune cells NFIL3 acts at lineage-specification loci rather than at metabolic promoters 24663216,25310240. On frequency: the AT-rich 8-mer is expected of order 10^5 times in 3.1 Gb of human sequence, so an isolated match carries little weight.
Biology and disease
Single knockouts of Dbp, Hlf or Tef in mice are mild, which is the expected signature of factors that share an element. Mice lacking all three are viable but highly susceptible to spontaneous and audiogenic generalized seizures that are frequently lethal; the mechanism was traced to reduced expression of pyridoxal kinase and hence lower brain pyridoxal phosphate, serotonin and dopamine 15175240. NFIL3 has non-redundant immune roles: it is required for commitment to the natural killer cell lineage, where it directly regulates Eomes and Id2 24663216, and for the development of a common innate lymphoid cell precursor 25310240. HLF is the 3-prime partner of the t(17;19) TCF3-HLF fusion in a rare and aggressive pro-B acute lymphoblastic leukemia, in which the HLF bZIP supplies dimerization and DNA binding to a chimeric activator 1386162. In normal human hematopoiesis, HLF expression defines the stem cell compartment 34499717.
Reading a hit
A TTACGTAA match is a clean family call by the standards of this class, because the family is cohesive, its models agree across databases, and several rest on in vitro binding data. It is not a member call at all, and here that matters more than usual: DBP, HLF and TEF activate while NFIL3 represses, so the same match is consistent with opposite outcomes, and which applies depends on cell type and on time of day. Check the outer bases before accepting the family. TTAC is PAR bZIP; TGAC is AP-1, CREB/ATF or ATF6/CREB3, and the shared ACGT core means a matrix scan will happily return CREs under a PAR model and vice versa. Two CIS-BP models filed here, for HLF and NFIL3, are themselves half-PAR and half-CRE hybrids. The 8-mer is AT-rich and common, so require accessibility or phase-resolved occupancy data before calling a site functional. Finally, treat a hit in a non-liver, non-immune, non-clock tissue with caution: expression, not sequence, is what limits this family's activity.
How this was worked out
DBP is cloned as the activator binding the D-site of the rat albumin promoter; it is liver-enriched at the protein level although its mRNA is widespread.
Still open then: The report described DBP as lacking a leucine zipper; its grouping with TEF and HLF as PAR bZIP proteins came later.
The t(17;19) translocation of pro-B acute lymphoblastic leukemia fuses the E2A activation domain to the bZIP of a new factor, HLF, closely related to DBP and TEF.
In vitro binding-site selection gives HLF and E2A-HLF the consensus GTTACGTAAT; HLF also binds related PAR and C/EBP sites, and the chimera is less tolerant of deviations from the consensus.
Still open then: Whether the TTAC half-site or the older TTATG(T/C)AA form better describes in vivo sites was not addressed.
- 1996mediumProtein quantification across the day, in vitro binding, co-transfection with domain-swap fusions8617210
TEF and DBP accumulate with strong circadian rhythms in liver and bind the same sequences in vitro, yet prefer different promoters; a TEF-DBP fusion carrying the DBP DNA-binding and dimerization domain behaves like DBP, so the preference does not track measured affinity.
Still open then: How the bZIP region confers promoter preference without changing in vitro affinity was not explained.
E4BP4/NFIL3 cycles in antiphase to DBP, HLF and TEF, cannot heterodimerize with them, and represses through the same sites they activate, making the family a clock-driven repressor-activator switch.
Mice lacking DBP, HLF and TEF develop frequently lethal epilepsy, traced to reduced pyridoxal kinase and lower brain pyridoxal phosphate, serotonin and dopamine.
- 2014mediumKnockout progenitor differentiation and rescue, clonal assays, binding to regulatory regions24663216 25310240
NFIL3 is required at the common lymphoid progenitor for NK-cell commitment, directly regulating Eomes and Id2, and for a committed precursor of all innate lymphoid cell lineages, acting through TOX.
HLF is one of the most selectively expressed genes in human hematopoietic stem cells, and an HLF reporter marks all stem-cell activity on serial transplantation.
Still unresolved
- Whether the PAR site is better written TTACGTAA or TTATG(T/C)AA is unsettled, and the split follows evidence type: HT-SELEX models for DBP, TEF, HLF and NFIL3 give the symmetric TTACGTAA, while the ChIP-derived HOCOMOCO models and JASPAR matrices for HLF and NFIL3 give TTATGTAAC or TTATGCAAC, and whether this reflects depletion of the CpG-containing palindrome in the genome, heterodimer partners in cells, or both, has not been resolved.
- Why DBP and TEF, which bind the same sequences in vitro, prefer different promoters is unexplained: in fusion experiments the preference follows the DBP DNA-binding and dimerization domain, yet in vitro affinity does not differ, which points to promoter context or partner interactions that have not been identified.
Of the eight bZIP families in this atlas, this is the one whose motif models agree best with each other and with in vitro binding data, and no member is flagged as belonging elsewhere. The limits are elsewhere. The site differs from a cyclic-AMP response element by only the outermost base of each half-site, so truncated or low-information matrices will not separate them, and two CIS-BP models here are already half-CRE hybrids. The 8-mer is AT-rich and occurs of order 100,000 times genome-wide. Most importantly, a hit does not indicate direction: DBP, HLF and TEF activate, NFIL3 represses, they occupy the same element in antiphase across the day, and sequence cannot distinguish them.
References
- Mueller CR, Maire P, Schibler U. DBP, a liver-enriched transcriptional activator, is expressed late in ontogeny and its tissue specificity is determined posttranscriptionally. Cell 1990;61:279-91. PMID 2331750
- Inaba T, Roberts WM, Shapiro LH, et al. Fusion of the leucine zipper gene HLF to the E2A gene in human acute B-lineage leukemia. Science 1992;257:531-4. PMID 1386162
- Hunger SP, Brown R, Cleary ML. DNA-binding and transcriptional regulatory properties of hepatic leukemia factor (HLF) and the t(17;19) acute lymphoblastic leukemia chimera E2A-HLF. Mol Cell Biol 1994;14:5986-96. PMID 8065331
- Fonjallaz P, Ossipow V, Wanner G, et al. The two PAR leucine zipper proteins, TEF and DBP, display similar circadian and tissue-specific expression, but have different target promoter preferences. EMBO J 1996;15:351-62. PMID 8617210
- Mitsui S, Yamaguchi S, Matsuo T, et al. Antagonistic role of E4BP4 and PAR proteins in the circadian oscillatory mechanism. Genes Dev 2001;15:995-1006. PMID 11316793
- Cowell IG. E4BP4/NFIL3, a PAR-related bZIP factor with many roles. Bioessays 2002;24:1023-9. PMID 12386933
- Gachon F, Fonjallaz P, Damiola F, et al. The loss of circadian PAR bZip transcription factors results in epilepsy. Genes Dev 2004;18:1397-412. PMID 15175240
- Reinke AW, Baek J, Ashenberg O, et al. Networks of bZIP protein-protein interactions diversified over a billion years of evolution. Science 2013;340:730-4. PMID 23661758
- Male V, Nisoli I, Kostrzewski T, et al. The transcription factor E4bp4/Nfil3 controls commitment to the NK lineage and directly regulates Eomes and Id2 expression. J Exp Med 2014;211:635-42. PMID 24663216
- Yu X, Wang Y, Deng M, et al. The basic leucine zipper transcription factor NFIL3 directs the development of a common innate lymphoid cell precursor. Elife 2014;3. PMID 25310240
- Lehnertz B, Chagraoui J, MacRae T, et al. HLF expression defines the human hematopoietic stem cell state. Blood 2021;138:2642-2654. PMID 34499717
15 motifs in this family — JASPAR 4 · HOCOMOCO 6 · CIS-BP 4 · Vierstra 1
| motif | database | consensus | bits |
|---|---|---|---|
| DBP.H14CORE.0.SM.BSMB | HOCOMOCO | GTTACGTAAt | 15.9 |
| NFIL3.H14CORE.1.S.BSB | HOCOMOCO | gTTACGTAAt | 15.2 |
| TEF.H14CORE.0.SM.BSMB | HOCOMOCO | GTTAcGTAA | 14.7 |
| M02849_2.00 | CIS-BP | gTTACGTAAt | 14.6 |
| TEF (MA0843.2) | JASPAR | aTTACGTAAc | 14.4 |
| DBP (MA0639.2) | JASPAR | aTTACGTAAc | 13.7 |
| HLF.H14CORE.0.P.BPB | HOCOMOCO | TTATGtAAc | 13.3 |
| HLF (MA0043.4) | JASPAR | TTATGCAAc | 13.1 |
| M04013_2.00 | CIS-BP | aTtACGTAAt | 13.1 |
| NFIL3.H14CORE.0.P.BPB | HOCOMOCO | gTTATGTAAc | 12.9 |
| NFIL3 (MA0025.3) | JASPAR | TTATGcAAt | 11.3 |
| HLF.H14CORE.1.SM.BSMB | HOCOMOCO | gTTAcGTAA | 11 |
| M04016_2.00 | CIS-BP | gTgACGTAA | 10.9 |
| AC0629:DBP/HLF:bZIP | Vierstra | TTAcGTAA | 10.8 |
| M04034_2.00 | CIS-BP | aTgACGTAAt | 9.7 |
HOCOMOCO tags read as assay + quality grade: P ChIP-seq, S HT-SELEX, M methyl-HT-SELEX, G GHT-SELEX, I SMiLE-seq, B PBM; grade A means two or more assay types agreed. A P-only matrix is a record of where the protein was crosslinked, which need not be a sequence it binds itself.
The factors, elsewhere
These entries are written from the primary literature and every citation is checked against PubMed, but this one can still be out of date, too confident, or simply wrong — especially where a motif’s published site and its database matrices disagree. Corrections and additions are welcome, and they are the fastest way to make this page better for everyone.