A conserved salt bridge in the G loop of multiple protein kinases is important for catalysis and for in vivo Lyn function

Abstract

The glycine-rich G loop controls ATP binding and phosphate transfer in protein kinases. Here we show that the functions of Src family and Abl protein tyrosine kinases require an electrostatic interaction between oppositely charged amino acids within their G loops that is conserved in multiple other phylogenetically distinct protein kinases, from plants to humans. By limiting G loop flexibility, it controls ATP binding, catalysis, and inhibition by ATP-competitive compounds such as Imatinib. In WeeB mice, mutational disruption of the interaction results in expression of a Lyn protein with reduced catalytic activity, and in perturbed B cell receptor signaling. Like Lyn(-/-) mice, WeeB mice show profound defects in B cell development and function and succumb to autoimmune glomerulonephritis. This demonstrates the physiological importance of the conserved G loop salt bridge and at the same time distinguishes the in vivo requirement for the Lyn kinase activity from other potential functions of the protein.

Fig. 1. Isolation of WeeB, a mouse mutant expressing catalytically inactive Lyn

(A) In WeeB mice, an A864G transition in Lyn exon 8 converts E₂₆₀ in the murine Lyn A kinase domain into a G (blue). E₂₆₀ is immediately C-terminal of the GxGxxG G-loop signature motif (last two Gs in orange). Nucleotide numbering is based on the mouse Lyn A cDNA sequence, Genbank Acc. No. M57696, amino acid numbering on AAA39471. (B) Immunoblot analysis of Lyn (upper) or Actin (lower panel) protein expression in splenic B cell extracts from C57BL/6 controls (wt), WeeB or Lyn^-/- mice at the indicated times after αIgM stimulation. LynA/B (Stanley et al., 1991; Yi et al., 1991) were not resolved and are represented by one band here and in fig. 5. Quantified band intensities are indicated. (C) Lyn was immunoprecipitated from splenic B cell extracts 1 min post αIgM stimulation, followed by Lance in vitro kinase assay. Representative of two independent experiments.

Fig. 2. Multiple phylogenetically diverse kinase domains contain a conserved salt bridge in their G-loop

(A) Amino acid sequence alignment of the G-loop region of all SFK family members (accession numbers in Supplemental Methods). h, human; m, murine; z, zebrafish; t, torpedo. β1, β2, β-strands flanking the G-loop. An extended SFK G-loop consensus sequence is shown below the alignment. Green, conserved Gs forming the characteristic protein kinase G-loop motif GxG₀xxGxV (Hanks and Hunter, 1995; Schenk and Snaar-Jagalska, 1999). Black, deviations from the consensus. Blue, a conserved basic residue (K/R) at position -4 N-terminally of the invariant G₀. Red, a conserved acidic E or D at position +4, corresponding to E₂₆₀ in murine Lyn A. (B) Top, ribbon diagrams of the kinase domain structures of human Hck (1AD5), Lck (2OF2), c-Src (1Y57), Abl (1IEP), zea mays CK2A1 (1LP4) and human SLK (2JFL) bound to ATP analogs. Orange, G-loop; colored, ATP-analog. Bottom, G-loop blowups showing electrostatic interactions between the conserved basic and acidic side chains flanking the LGxG₀xFG motif at positions -4 and +4. The polarity of the SLK interaction is reversed (Supplemental fig. 2). (C) Phylogenetic tree of all eukaryotic protein kinase domains from kinase.com/human/kinome/phylogeny.html, rendered with HyperTree. Red dots indicate the kinases in (B), listing major family_subfamily_member. TK, tyrosine kinase; Src, Src family; STE20, Ste20-family kinase; Other, no major group. (D) Disruption of the conserved SFK G-loop salt bridge abrogates catalytic activity. Recombinant full-length mouse Lyn or the G-loop mutants E₂₆₀G (Lyn^WeeB), K₂₅₂G, K₂₅₂G/E₂₆₀G, K₂₅₂E/E₂₆₀K or K₂₅₂E were expressed in Sf21 cells to the levels shown via IB (upper panel). Lower panel, catalytic activity in vitro. Error bars, standard deviations of triplicates. Representative of 3 independent experiments.

Fig. 3. Disruption of the conserved salt bridge destabilizes the SFK G-loop and enhances its flexibility

(A) Blowup of the G-loop region (dark blue) of mLyn, homology modeled into the active Hck/AMP-PNP kinase domain structure (Fig. 2B), showing the side chains of the conserved basic K_-4 (K₂₅₂) and acidic E₊₄ (E₂₆₀). Juxtaposition at 2.3 Å suggests a salt bridge interaction. (B) Molecular Dynamics simulation of the flexibility of the modeled wt (red) or WeeB mutant (E₂₆₀G, blue) Lyn G-loop. Shown is the root mean square displacement (RMSD) over time. (C) Simulated atomic fluctuations for amino acids around the catalytic site in the wt (red) or WeeB mutant (blue) Lyn model. (D) Superimposed average structures of the wt (red) and WeeB mutant (blue) Lyn models from a 5 ns molecular dynamics simulation. The position of the docked ATP is shown for reference.

Fig. 4. Perturbed splenic B cell development in WeeB mice

Bar graph depicting mean numbers ± standard deviation of total B220⁺, follicular mature (FM, B220⁺IgM^lowCD21⁺), marginal zone (MZ, B220⁺IgM^highCD21⁺CD23^-), transitional T2 (B220⁺IgM^highCD21⁺CD23⁺) and T1 (B220⁺IgM^highCD21^-) B cells in wt, WeeB or Lyn^-/- mice. Average live cell numbers in 3 wt, 4 WeeB or 2 Lyn^-/- spleens were 5.6±0.7 ×10⁷, 2.3±0.7 ×10⁷ and 1.5±0.1 ×10⁷, respectively. Representative FACS data are shown in supplemental fig. 7.

Fig. 5. Impairment of Lyn signaling downstream of the BCR in WeeB mice

Primary B cells from 7-11 m old wt, WeeB and Lyn^-/- mice were stimulated with 10 μg/ml αIgM F(ab')₂ for the indicated times in min. (A, C, D) Lysates (25 μg total protein each) were resolved via SDS-PAGE and analyzed by IB with ABs against phospho-tyrosine (4G10, A) or the indicated proteins. P-, phosphorylated protein epitope; Actin, loading control. (B, E) Lyn (B) or CD22 (E) were immunoprecipitated (IP) from lysates (500 μg protein) and analyzed via IB with ABs against Lyn (B), phospho-CD22 (Y₇₆₂, Ty2, E), or SHP-1 after stripping (E). WCL, whole cell lysate controls. The numbers underneath the panels indicate actin-normalized band intensities as fold change over the corresponding unstimulated 0 min sample. Representative of 3 independent experiments.

Fig. 6. Perturbed B cell function and late-onset autoimmune disease in WeeB mice

(A) Splenic B cells were incubated with the indicated concentrations of αIgM F(ab')₂ for 3 days. Proliferation was determined via ³H-Thymidine incorporation. Shown are means ± standard deviation (n=3). (B) Enlarged spleen size in 9 m old Lyn^-/- or 12 m old WeeB mice compared to 12 m old wt controls. (C) Serum α-dsDNA AB levels in 3 or 10 months (m) or 1 year old littermate controls (LMC), WeeB or Lyn^-/- mice. Shown are mean OD₄₅₀s of duplicate measurements, each bar represents one mouse. (D) Histological analyses of renal cortex, liver or spleen sections from C57BL/6 controls, WeeB or Lyn^-/- mice at the indicated ages. H&E, Hematoxylin/Eosin staining. PAS-M, periodic acid-methananine silver staining. IgG, α-IgG, B220/CD3, B220 (brown, B cells)/CD3 (blue, T cells) staining. All animals and phenotypes are summarized in supplemental tables 2,3.