KDM5A

Histone lysine demethylase 5A (KDM5A) from Homo sapien (PBD ID: 5CEH) is a lysine specific demethylase that catalyzes the demethylation of Lys-4 on Histone 3 (1). The demethylase is an important part of transcription regulation in cell development and differentiation (2). Lysine demethylases are enzymes that catalyze the removal of methyl groups and have recently become promising cancer drug targets because of demethylase necessity in regulating cancer cell drug tolerance (1, 3). The structure and mechanism by which KDM5A acts as a demethylase is of great biological importance, as understanding tolerance could lead to drug resistance prevention therapy development opportunities (3).

The molecular weight of the 790 residue long KDM5A is 90,590.92 Da and its theoretical isoelectric point (pI) calculated from ExPASy, a bioinformatics database that calculates protein parameters, is 5.61 (4, 5). There are no classified alternative conformations, however, KDM5A is the A chain of KDM5, and is directly involved in the histone demethylation (5). A short B chain of unknown peptide was also crystallized, but is thought to be an unclassified part of chain A since a large portion of chain A is missing in the primary structure (5). Histone demethylases usually contain multiple domains for substrate binding, but KDM5A is interesting in that the Jumonji (Jmj) domain is split into the Jumonji N and catalytic Jumonji C domains by the insertion of two separate domains (6). The two Jumonji domains are two non-adjacent domains which have been identified in the Jumonji family of transcription factors (6). The inserted AT-rich interacting (ARID) and plant homeodomain 1 (PHD1) domains may block the enzyme active site (1). This may affect substrate binding, which also shows the involvement of PHD1 in the regulation of catalytic activity (2). The enzyme is of right hand topology, with the active site in the “palm” of the JmjC domain and the “thumb” formed by the zinc-binding module (1).

KDM5A secondary structure has primarily been identified and explained by comparing the structure to that of KDM6A (PBD ID: 3AVS). This protein, otherwise known as UTX, is another lysine specific demethylase also found to be important in development and acts as a tumor suppressor (1). An important element of the secondary structure of KDM5A is the alpha helix near the JmjC domain, which is tilted 26 degrees further away from the palm than KDM6A at residue 638 (1). CS-Rosetta Model calculations of the KDM5A PHD1 domain, which calculates biological macromolecules on the basis of conformational information from NMR, showed that the secondary structure is well-defined with a core beta-sheet and a C-terminal alpha helix (2). There is also a zinc binding site behind the B-sheet which is formed from the unwinding and shortening of the H1 helix in KDM6A by 4 amino acids (1). This coordinates the N terminal of the substrate at Histone 3, which is the part that acts as the demethylase for the lysine (1). There are several amino acid differences between KDM5A and KDM6A important in the binding of the zinc ligand, in which the KDM5A has three cysteines and a histidine instead of the four cysteines found in KDM6A (1). This results in the KDM5A creating a deeper cleft at the peptide binding junction and acting on the substrate closer to the N terminus (1). Because of this, the KDM5A would not be able to use the same peptide binding mechanism as KDM6A and so it is unclear whether the KDM5A zinc module is involved in protein binding of the substrate peptide or in the binding to other interacting proteins (1).

CPI-455 is a selective inhibitor of KDM5A that modulates activity in cells by elevating levels of histone methylation (1). This inhibitor is of great importance because in several experiments, demethylase activity was found to be involved in drug tolerance, as CPI-455 treatment led to a reduced number of drug tolerant persistent cancer cells (1). The structure of KDM5A is important in the inhibitor’s mechanism and selectivity, so understanding the structure could lead to potential clinical relevance (1). The nitrile of CPI-455 interacts with the active site nickel ion in the complex formation (1). In the binding pocket, the 7th position carbonyl oxygen of the CPI-455-metal ion complex hydrogen bonds with the side chain Asn-575 of KDM5A (Fig. 1) (1). The aromatic core forms pi stacks with the side chain Tyr-472 and Phe-480, as well as an aromatic contact with Trp-503 (1). Furthermore, the Lys-501 N terminal is positioned 3.6Å from CPI-455, meaning no hydrogen bonding can occur (1). All amino acids within 4Å of the CPI-455 are conserved among all KDM families (1). The 6-isopropyl of the CPI-455 tightly packs in the cleft formed by Tyr-409 and Ser-478, overlapping other binding sites, which shows a competitive mode of binding (1).

KDM5A is also known as Retinoblastoma protein 2 (RBP-2) as it was first identified as a nuclear phosphoprotein, a potential retinoblastoma binding protein (7). However, Klose et al. identified demethylation properties of the protein, including the display of robust histone demethylase activity against the me3 and me2 modification states, resulting in over 80% demethylation of the modified substrate (7). However, the protein could not catalyze the removal of the me1 modification state in vitro (7). This enzymatic property of being able to catalyze the removal of three modification states, but not being able to initiate the me1 modification in vitro, was found to be similar to that of other trimethyl demethylases (7). These findings suggest that certain methyl modification states are enzymatically reversible and KDM5A/RBP-2 is involved in the histone demethylation contribution to transcriptional regulation (7). Transcription regulation of KDM5A involves DNA binding transcription factors and the ARID domain (8). The tertiary structure of the ARID domain is similar to that found in other proteins, consisting of six helices and two loops, in which three helices form a U shape (8). The domain structure shows the preferential binding of the KDM5A/RBP-2 ARID domain to GC motifs on DNA (8).

There are many family members in the lysine demethylases, including the third Y chromosome linked male-specific N epsilon-methyl lysyl demethylase KDM6C, otherwise known as UTY, from Spodoptera frugiperda (PBD ID: 3ZLI) (5, 9). Position-Specific Iterated Basic Local Assignment Search Tool (PSI-BLAST), an algorithm based protein database that looks at primary sequence alignment, indicated significant homology (10). E-values less than 0.05 represent high homology and low gaps in nonhomologous areas when primary sequences are compared (10). Gaps indicates amino acids that exist in the subject protein, the protein with the similar primary structure, but not in the query protein (10). An E-value of 0.04 was given between KDM5A and KDM6C indicating high homology between the two proteins without being identical (10). The Dali server was also used to compare tertiary structures of the two proteins using the sums-of-pairs method and calculating the differences in intramolecular distances (11). UTY also had a high Z score (Z score > 2) with KDM5A, indicating high homology and similar folds (Z score = 25.4) (11).

KDM6C/UTY shares more than 88% structural similarity with KDM6A/UTX, the X chromosome homologue with structure previously described to be similar to KDM5A (9). However, it was reported to have significantly decreased lysine demethylase activity (9). Because down-regulating UTY activity is associated with increased risk of male cardiovascular disease, understanding the role of UTY is important in both a scientific and therapeutic perspective (9). KDM6C/UTY shares many similar structures with KDM5A and KDM6A/UTX including the three conserved domains- JmjC, linker helical and zinc binding (9). Furthermore, the secondary and tertiary folds are almost identical between KDM6C/UTY and KDM6A/UTX (9). The KDM6C/UTY JmjC domain consists of 13 B strands and 10 helices, similar to other proteins in the 2-oxoglutarate oxygenase superfamily, which includes KDM5A (9).

However, the linker region between the helix and JmjC domains were not observed in KDM6C/UTY (9). Several residues were found to be important in the reduction of lysine demethylase activity in the UTY (9). An Ile-1267 in KDM6A involved in hydrophobic interactions with Ala-25 of histone lysines is substituted for a Pro-1214 in UTY (9). Sequence alignments also found Ser-1138 and Pro-1214 in place of Gly and Ile, which could influence peptide binding and result in decreased activity (9). The KDM activity was shown to be rescued by substitution of Pro-1214 with Ile to promote substrate binding (9). Mutations made in KDM6A/UTX mice also found that Tyr-1135 is conserved in histone lysine demethylases, and the Cys-947 found in UTY mice would not effectively interact with the methyl groups of Histone 3 (12). Furthermore, a Thr-1143 conserved in the demethylases was also replaced by a bulky isoleucine, which would remove the interaction with the hydroxyl group as well as sterically hinder the binding site, leading to reduced catalytic activity (12). Crystal structure of the JmjC and zinc binding domains showed identical binding to 2-oxogutarate and inhibitor GSK-J1, indicating that UTY is a catalytically functional KDM despite the low KDM activity (9).

Histone lysine methylation pathways are important targets for cancer treatment drug discovery (1). Although JmjC type and 2-OG dependent demethylases are difficult to specifically inhibit, the complete structure of the multidomain lysine demethylase KDM5A with the selective inhibitor CPI-455 shows the capacity for in vitro exploration of KDM5 dependent mechanisms as well as drug tolerance in cancer cells (1). Future studies could lead to discovery of more inhibitors as well as expanded studies of in vivo experiments for successful cancer therapy (1).