PRDM
PRDM proteins are zinc-finger transcription factors defined by an N-terminal PR domain, a variant of the SET methyltransferase fold, followed by a tandem array of C2H2 zinc fingers that does the DNA binding. The family is heterogeneous and does not share one motif: each member's finger array reads its own sequence, and those sequences are unrelated to one another. The best-characterized members are PRDM1/BLIMP1, which binds an interferon-regulatory-factor-like AAAGTGAAAG site and drives plasma-cell differentiation; PRDM9, whose rapidly evolving finger array selects meiotic recombination hotspots and underlies a mouse hybrid-sterility phenotype; PRDM14, in naive pluripotency; and PRDM16, in brown adipocyte and craniofacial development. A hit labelled PRDM should be interpreted per member, never as a family-level call.
The factors
The PRDM proteins take their name from the PR domain, a diverged SET domain of about 130 residues, followed by a variable number of C2H2 zinc fingers 22669819. Humans have roughly 19 PRDM genes. This family groups the ones for which a motif has actually been measured: PRDM1, PRDM4, PRDM5, PRDM6, PRDM9, PRDM10, PRDM13, PRDM14, PRDM15 and PRDM16. MECOM (historically PRDM3) is kept as a separate family here because its two zinc-finger clusters behave differently from a single array.
Only some PR domains are catalytically active. PRDM9 is a genuine histone H3 lysine-4 trimethyltransferase, first described in mouse as Meisetz, and that activity is required for meiotic prophase 16292313. For most other members the PR domain has no demonstrated methyltransferase activity and acts as a protein-interaction surface; PRDM1/BLIMP1, for instance, works by recruiting histone deacetylase and methyltransferase cofactors rather than by modifying histones itself 22669819. Expression is mostly restricted. PRDM1 appears in plasma cells, effector T cells, keratinocytes, primordial germ cells and placenta; PRDM9 essentially only in meiotic germ cells; PRDM14 in embryonic stem cells and germ cells; PRDM16 in brown and beige adipocytes, craniofacial mesenchyme and hematopoietic stem cells. Few PRDMs are broadly expressed, so the plausible member behind a hit is usually decided by cell type rather than by sequence.
The motif
There is no family consensus, and the data in this atlas show that plainly: member motifs range from the GTTTCAAG core of PRDM4, through the IRF-like AAAGTGAAAG of PRDM1, to GGTCTCTAAC for PRDM14 and G-rich models for PRDM9 and PRDM16. Cohesion across the family is low, about 0.5, and the medoid motif (ACTTTCACT, the PRDM1 site on the reverse strand) is simply the most central member, not a shared element.
PRDM1/BLIMP1 is the clearest case. Its zinc fingers read an approximately 10-bp site built on a GAAAG core; the optimal selected sequence is close to AAAGTGAAAG, which overlaps a subset of interferon-stimulated response elements, so BLIMP1 and IRF1/IRF2 compete for the same DNA with comparable affinities 15494505. In vivo, several motif variants are needed to capture BLIMP1 occupancy, and binding is sensitive to CpG methylation 20421211. The reverse-complement rendering ACTTTCACTTT is the same site and is what several database entries display.
PRDM9 is different in kind. Its DNA-binding module is a minisatellite-encoded tandem array of roughly a dozen C2H2 fingers, one of the fastest-evolving coding sequences in mammals, polymorphic within humans and between mouse strains 20044539. Human hotspots are enriched for a degenerate 13-mer, CCNCCNTNNCCNC, which the common human A allele recognizes 19165926,20044539; crystal structures of the human PRDM9 finger array on DNA explain the base contacts, and the C allele reads a different sequence 26833727. Because the array differs between alleles, no single PRDM9 matrix is correct for all individuals, and the JASPAR and CIS-BP PRDM9 models are approximations of one allele's preference at best.
Where it acts
PRDM1 sites sit in promoters of genes it represses, including IRF-dependent immune programs, in both CpG-poor and CpG-island promoters, with the methylation-sensitive variants biased to the latter 20421211. PRDM9 sites are the defining feature of meiotic double-strand-break hotspots: tens of thousands of them, typically 1-2 kb wide, mostly outside promoters, and in mouse directed away from functional elements such as transcription start sites. PRDM14 and PRDM15 sites are concentrated at enhancers and promoters of pluripotency and signaling genes in stem cells. PRDM16 sites are found at brown-fat enhancers, where PRDM16 acts less as an autonomous sequence reader than as a partner of adipogenic factors 18719582, which matters for interpreting its ChIP-derived motif.
Biology and disease
PRDM1 knockout blocks plasma-cell formation and specification of primordial germ cells; in humans PRDM1 is inactivated by mutation and deletion in diffuse large B-cell lymphoma, where it behaves as a tumor suppressor 16492805. PRDM9-null mice are sterile with arrested meiosis 16292313; PRDM9 allele identity changes hotspot usage in human families 20044539; and one PRDM9 allele is the mouse hybrid-sterility gene, with sterility reversible by swapping the finger array 26840484. PRDM14 loss destabilizes naive pluripotency and PRDM15 loss does the same through WNT and MAPK-ERK targets. PRDM16 loss impairs brown-fat identity 18719582, and PRDM16 mutations found in nonsyndromic left ventricular noncompaction and dilated cardiomyopathy were proposed to cause the cardiomyopathy of 1p36 deletion syndrome 23768516, an attribution others consider unlikely 24387995; PRDM5 loss-of-function causes brittle cornea syndrome 21664999; PRDM6 variants cause nonsyndromic patent ductus arteriosus 27181681. PRDM10 and PRDM16 also appear as fusion partners in soft-tissue tumors and leukemia.
Reading a hit
Treat a PRDM hit as a hit on one named member. Sequence cannot tell you which PRDM bound, because the members do not share a motif and the family label spans unrelated sites. Check which member the underlying matrix is labelled with, then whether that member is expressed in your cell type. For PRDM1 the site is short and IRF-like, so an IRF or ISRE explanation is at least as likely as BLIMP1 unless BLIMP1 is present 15494505,20421211. For PRDM9 a match is meaningless outside meiotic germ cells, and allele-dependent even there. Several PRDM entries rest on ChIP-only evidence, where the motif may belong to a partner; the EBF-like HOCOMOCO PRDM16 model (PRD16.H14CORE.0.P.B) is one. The HOCOMOCO PRDM13 model, which matches a SNAI-type CAGGTG E-box, is not: PRD13.H14CORE.0.PSGI.A adds HT-SELEX, GHT-SELEX and SMiLE-seq to ChIP-seq at grade A, so indirect binding does not explain it. Both are flagged as inconsistent with their labels by their nearest neighbours across databases.
How this was worked out
PRDI-BF1, a five-zinc-finger protein, is identified as a repressor that binds the PRDI element of the human interferon-beta promoter and is proposed to act by displacing its activators.
Still open then: The binding site was known only as PRDI; its relation to IRF sites and to B-cell differentiation was not yet recognized.
Mouse Blimp-1, the orthologue of PRDI-BF1, is cloned as a zinc-finger factor induced in, and able to drive, the maturation of B cells into immunoglobulin-secreting cells.
Still open then: Its DNA-binding specificity had not been defined.
- 2004highIn vitro site selection; EMSA competition and dissociation constants; ChIP; cotransfection15494505
The optimal BLIMP1 site is determined and found to match the subset of IRF sites containing GAAAG; BLIMP1, IRF1 and IRF2 bind such sites with similar affinities and compete at the IFNB promoter.
Still open then: How many genomic BLIMP1 sites are shared with IRFs was unknown.
Meisetz (PRDM9) is shown to be a histone H3 lysine-4 trimethyltransferase expressed only in germ cells entering meiotic prophase; knockout mice of both sexes are sterile.
Still open then: What targets PRDM9 to particular sites in the genome was not addressed.
A degenerate 13-mer, CCNCCNTNNCCNC, is found to recruit crossovers to at least 40% of human recombination hotspots.
Still open then: The protein that reads the 13-mer was unknown.
- 2010highMouse strain genetics; zinc-finger recognition prediction; in vitro binding; human association20044539
PRDM9 is identified as the major determinant of hotspot location: the human consensus allele is predicted, and shown in vitro, to bind the 13-mer, and PRDM9 finger-array alleles track hotspot usage in mice and humans.
Still open then: Because the finger array is highly polymorphic, a single matrix could describe only one allele.
Genome-wide BLIMP1 promoter occupancy shows that several motif variants are needed to capture its specificity, that binding is methylation-sensitive, and that only a subset of occupied sites overlap IRF motifs.
- 2016highX-ray crystallography; in vitro binding of allelic variants; humanized-allele mouse genetics26833727 26840484
A crystal structure of PRDM9 allele A fingers 8-12 on hotspot DNA explains its base contacts and allele-specific affinities, and humanizing the mouse PRDM9 finger array repositions hotspots and restores fertility in sterile hybrids.
Still open then: Why symmetric binding of both homologues is needed for fertility remains mechanistically unexplained.
Still unresolved
- Whether PRDM16 loss causes the cardiomyopathy of 1p36 deletion syndrome is disputed: Arndt et al. (2013) mapped a minimal deletion to PRDM16 and found PRDM16 mutations in noncompaction and dilated cardiomyopathy, while de Leeuw and Houge (2014) argued that PRDM16 loss is unlikely to explain the cardiomyopathy in the deletion syndrome.
- The HOCOMOCO PRDM13 model (GCAGGTGG) looks like a SNAI-type E-box, yet it carries in vitro support (HT-SELEX, GHT-SELEX and SMiLE-seq, grade A), so it is unresolved whether PRDM13 genuinely reads an E-box-like site or the model is mislabelled; the EBF-like PRDM16 model, by contrast, rests on ChIP-seq alone, where a partner motif is the simpler explanation.
- How PRDM9 binding leads to fertile meiosis is not settled: hotspot position follows the allele's binding specificity, but in hybrid mice fertility tracks whether PRDM9 binds both homologues symmetrically, and the downstream step that requires symmetry is unknown.
- Which BLIMP1 sites are direct high-affinity targets and which are competitive IRF-overlapping sites is only partly resolved: in vitro selection gives one AAAGTGAAAG-like site, while promoter occupancy needs several motif variants that differ in CpG content and methylation sensitivity.
This family has no shared motif, so a family-level hit means little; only the member-specific matrix matters. Cohesion across members is about 0.5, and two entries contradict their labels: the HOCOMOCO PRDM16 model (TCCCCTGGGA) resembles an EBF site and rests on ChIP-seq alone, where a partner motif is the simpler explanation, while the PRDM13 model (GCAGGTGG) is a SNAI-type E-box despite grade-A in vitro support, so it is either a genuine E-box-like preference or a mislabel. PRDM1 evidence is strong and includes in vitro selection, but the site overlaps IRF elements, so IRF1, IRF2 or an ISRE is a competing interpretation. PRDM9 matrices describe one allele of a polymorphic finger array and are interpretable only in meiotic germ cells. Short G-rich PRDM models match frequently by chance.
References
- Turner CA Jr, Mack DH, Davis MM. Blimp-1, a novel zinc finger-containing protein that can drive the maturation of B lymphocytes into immunoglobulin-secreting cells. Cell 1994;77:297-306. PMID 8168136
- Keller AD, Maniatis T. Identification and characterization of a novel repressor of beta-interferon gene expression. Genes Dev 1991;5:868-79. PMID 1851123
- Kuo TC, Calame KL. B lymphocyte-induced maturation protein (Blimp)-1, IFN regulatory factor (IRF)-1, and IRF-2 can bind to the same regulatory sites. J Immunol 2004;173:5556-63. PMID 15494505
- Doody GM, Care MA, Burgoyne NJ, et al. An extended set of PRDM1/BLIMP1 target genes links binding motif type to dynamic repression. Nucleic Acids Res 2010;38:5336-50. PMID 20421211
- Pasqualucci L, Compagno M, Houldsworth J, et al. Inactivation of the PRDM1/BLIMP1 gene in diffuse large B cell lymphoma. J Exp Med 2006;203:311-7. PMID 16492805
- Hayashi K, Yoshida K, Matsui Y. A histone H3 methyltransferase controls epigenetic events required for meiotic prophase. Nature 2005;438:374-8. PMID 16292313
- Myers S, Freeman C, Auton A, et al. A common sequence motif associated with recombination hot spots and genome instability in humans. Nat Genet 2008;40:1124-9. PMID 19165926
- Baudat F, Buard J, Grey C, et al. PRDM9 is a major determinant of meiotic recombination hotspots in humans and mice. Science 2010;327:836-40. PMID 20044539
- Patel A, Horton JR, Wilson GG, et al. Structural basis for human PRDM9 action at recombination hot spots. Genes Dev 2016;30:257-65. PMID 26833727
- Davies B, Hatton E, Altemose N, et al. Re-engineering the zinc fingers of PRDM9 reverses hybrid sterility in mice. Nature 2016;530:171-176. PMID 26840484
- Seale P, Bjork B, Yang W, et al. PRDM16 controls a brown fat/skeletal muscle switch. Nature 2008;454:961-7. PMID 18719582
- Hohenauer T, Moore AW. The Prdm family: expanding roles in stem cells and development. Development 2012;139:2267-82. PMID 22669819
- Burkitt Wright EMM, Spencer HL, Daly SB, et al. Mutations in PRDM5 in brittle cornea syndrome identify a pathway regulating extracellular matrix development and maintenance. Am J Hum Genet 2011;88:767-777. PMID 21664999
- Arndt AK, Schafer S, Drenckhahn JD, et al. Fine mapping of the 1p36 deletion syndrome identifies mutation of PRDM16 as a cause of cardiomyopathy. Am J Hum Genet 2013;93:67-77. PMID 23768516
- de Leeuw N, Houge G. Loss of PRDM16 is unlikely to cause cardiomyopathy in 1p36 deletion syndrome. Am J Hum Genet 2014;94:153-4. PMID 24387995
- Li N, Subrahmanyan L, Smith E, et al. Mutations in the Histone Modifier PRDM6 Are Associated with Isolated Nonsyndromic Patent Ductus Arteriosus. Am J Hum Genet 2016;98:1082-1091. PMID 27181681
27 motifs in this family — JASPAR 6 · HOCOMOCO 11 · CIS-BP 4 · Vierstra 6
| motif | database | consensus | bits |
|---|---|---|---|
| M04421_2.00 | CIS-BP | aaCaGTTTCAAGgCcCCCC | 23.3 |
| AC0287:PRDM:C2H2_ZF | Vierstra | gCrGTTTCAAGGCtcCC | 19.9 |
| PRDM5.H14CORE.0.PSGI.APSGIA | HOCOMOCO | AcCctGatCaCCgtbcGnnCCtt | 16.4 |
| M08277_2.00 | CIS-BP | TTTTttTbtTTTTTthCTtT | 16.3 |
| M07610_2.00 | CIS-BP | TtcTdnbnCCcCCcTnsCCnC | 16.2 |
| PRDM9 (MA1723.2) | JASPAR | GtGGgvAgGGaGGnngcAg | 15.6 |
| PRDM1.H14CORE.1.PS.APSA | HOCOMOCO | AAAGTGAAAGT | 15 |
| M07845_2.00 | CIS-BP | ACTTTCACTTT | 15 |
| AC0451:PRDM:C2H2_ZF | Vierstra | aTCACTTT | 15 |
| PRDM9.H14CORE.0.P.BPB | HOCOMOCO | AdadGGcAGcAgCAyCt | 14.7 |
| PRD10.H14CORE.0.PSGI.APSGIA | HOCOMOCO | aTGGTAcagtCCa | 14.3 |
| PRDM4.H14CORE.0.PSM.APSMA | HOCOMOCO | ggcaGTTTCaAGgnhcCC | 13.9 |
| Prdm4 (MA1647.3) | JASPAR | CCTTGAAACtG | 13.8 |
| PRDM1.H14CORE.0.S.BSB | HOCOMOCO | aAGTGAAAGTg | 13.5 |
| Prdm15 (MA1616.2) | JASPAR | gAAAACCTGGA | 13 |
| PRD15.H14CORE.0.P.BPB | HOCOMOCO | AAAACCtGGavngG | 11.7 |
| AC0143:PRDM:C2H2_ZF | Vierstra | GGTCTCTAAC | 11.6 |
| Prdm5 (MA1999.2) | JASPAR | GTTCTCCATCt | 11.4 |
| PRDM1 (MA0508.4) | JASPAR | CTTTCTC | 11.2 |
| PRD13.H14CORE.0.PSGI.APSGIAlooks like SNAI | HOCOMOCO | GCaGGTGg | 11.1 |
| AC0674:PRDM:C2H2_ZF | Vierstra | ACTTTCaCT | 10.9 |
| PRD16.H14CORE.0.P.BPBlooks like EBF | HOCOMOCO | TCcCCTGGGA | 10.6 |
| Prdm14 (MA1998.2) | JASPAR | GGTCTCTA | 10.5 |
| PRD14.H14CORE.0.P.BPB | HOCOMOCO | gTTAGAGACC | 10.3 |
| AC0220:PRDM:C2H2_ZF | Vierstra | TTTTTCtT | 9.8 |
| PRDM6.H14CORE.0.P.BPB | HOCOMOCO | aGAAAA | 9.5 |
| AC0288:PRDM/ZEB:['C2H2_zinc_finger_factors'] | Vierstra | GAAACaG | 8.9 |
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.