HPK1 Associates with SKAP-HOM to Negatively Regulate Rap1-Mediated B-Lymphocyte Adhesion

Background Hematopoietic progenitor kinase 1 (HPK1) is a Ste20-related serine/threonine kinase activated by a range of environmental stimuli including genotoxic stress, growth factors, inflammatory cytokines and antigen receptor triggering. Being inducibly recruited to membrane-proximal signalling scaffolds to regulate NFAT, AP-1 and NFκB-mediated gene transcription in T-cells, the function of HPK1 in B-cells to date remains rather ill-defined. Methodology/Principal Findings By using two loss of function models, we show that HPK1 displays a novel function in regulating B-cell integrin activity. Wehi 231 lymphoma cells lacking HPK1 after shRNA mediated knockdown exhibit increased basic activation levels of Ras-related protein 1 (Rap1), accompanied by a severe lymphocyte function-associated antigen-1 (LFA-1) dependent homotypic aggregation and increased adhesion to intercellular adhesion molecule 1 (ICAM-1). The observed phenotype of enhanced integrin activity is caused downstream of Src, by a signalling module independent of PI3K and PLC, involving HPK1, SKAP55 homologue (SKAP-HOM) and Rap1-GTP-interacting adaptor molecule (RIAM). This alters actin dynamics and renders focal adhesion kinase (FAK) constitutively phosphorylated. Bone marrow and splenic B-cell development of HPK1−/− mice are largely unaffected, except age-related tendencies for increased splenic cellularity and BCR downregulation. In addition, naïve splenic knockout B-cells appear hyperresponsive to a range of stimuli applied ex vivo as recently demonstrated by others for T-cells. Conclusions/Significance We therefore conclude that HPK1 exhibits a dual function in B-cells by negatively regulating integrin activity and controlling cellular activation, which makes it an interesting candidate to study in pathological settings like autoimmunity and cancer.


Introduction
Integrin regulation is essential in mediating homeostatic lymphocyte features like adhesion, migration and immune synapse formation [1][2][3]. While being kept in an inactive state in resting cells, inside-out activation signals from the BCR, TCR or chemokine receptors facilitate the binding of the b integrin chain to talin [4]. As a consequence, membrane distribution and/or affinity for ligands inducibly change and concomitant outside-in integrin signalling can lower antigenic activation thresholds for B-and T-cells [5,6]. Recent studies propose B-cells to mainly recognize antigen in the form of immune complexes bound to Fc or complement receptors on e.g. follicular dendritic cells [7]. If the signals provided in the context of this interaction exceed a certain activation threshold, arrest of the Bcell eventually allows the formation of a peripheral supramolecular activation cluster composed of lymphocyte function-associated antigen-1 (LFA-1) and very late antigen-4 (VLA-4) integrins, which subsequently serve as a docking platform to increase the duration of the synapse and ensure proper B-cell activation [6,8].
The small GTPase Ras-related protein 1 (Rap1) belongs to the Ras superfamily and has a pivotal function in lymphocyte integrin activation [9][10][11]. By binding downstream effectors like regulator for cell adhesion and polarization enriched in lymphoid tissues (RAPL) and Rap1-GTP-interacting adaptor molecule (RIAM), Rap1 connects to talin [12] and induces lymphocyte polarization by distributing LFA-1 to the leading edge, which is indispensable for cellular adhesion and migration [13,14]. In T-cells, the two constitutively associated adaptor molecules adhesion-and degranulation-promoting adaptor protein (ADAP) and Src-kinase-associated phosphoprotein of 55 kDa (SKAP55) were shown to be of critical importance in translating signals from the TCR via Rap1 activation to LFA-1 and VLA-4 integrins [15][16][17][18][19]. Although ADAP and SKAP55 are not expressed in B-cells, a SKAP55 homologue (SKAP-HOM) broadly expressed in the hematopoietic system fulfils similar tasks in B-cells and macrophages [20][21][22].
In the current paper we demonstrate that HPK1, a Ste20-related serine/threonine kinase triggering the SAPK/JNK pathway [23,24], constitutively associates with SKAP-HOM in Wehi 231 cells, forming a ternary complex of HPK1, SKAP-HOM and RIAM. Wehi 231 cells lacking HPK1 expression after shRNA mediated knockdown display substantially increased LFA-1 mediated homotypic aggregation and adhesion to ICAM-1. This behaviour is caused by an upregulation of Rap1-GTP in unstimulated cells, leading to dysregulated actin dynamics and enhanced focal adhesion kinase (FAK) activity. We further show that HPK1 2/2 mice develop normal bone marrow and splenic B-cell subset counts but exhibit an increase in B-cell reactivity, suggesting that HPK1 limits the actions emanating from the SKAP-HOM adaptor complex needed for proper B-cell activation and adhesion.

HPK1 negatively regulates B-cell adhesion
For the analysis of HPK1 deficiency in B-cells the immature Bcell line Wehi 231 was chosen as target for RNA interference. shRNA mediated silencing reduced HPK1 protein expression by approximately 95% ( Figure 1A). While control (co) target cells showed a wild type like growth behaviour of single cells and small homotypic aggregates ( Figure 1B, upper left), Wehi 231 HPK1 knockdowns (kd) formed large cellular clusters ( Figure 1B, lower left) phenocopying the general growth of proliferating primary Bcells in vitro. As these cellular interactions are critically dependent on integrins, we speculated about an impact of HPK1 on integrin regulation. Blocking a L b 2 integrin with an anti-LFA-1 specific antibody applied to the medium for 17 h completely interfered with co and kd cell aggregation ( Figure 1B, right), suggesting that increased cellular clustering was caused by augmented LFA-1 activity, whereas the lack of HPK1 had no influence on total membrane LFA-1 expression as measured by flow cytometry ( Figure 1C). To further monitor the adhesion behaviour of transfectants under conditions of manually applied shear stress, co and kd cells were subjected to static adhesion on ICAM-1/Fc coated plates. Wehi 231 cells lacking HPK1 exhibited a twofold increase (p,0.001) in adhesion to ICAM-1/Fc compared to co cells ( Figure 2B) and showed a significantly higher percentage (p,0.001) of spreading cells (Figure 2A). Inside-out activation of integrins triggered by the BCR was shown to be regulated by the sequential activation of signalling mediators including Src family kinases, PI3K, PLCc2, Btk, guanosine nucleotide exchange factors like e.g. Vav1/2 as well as small GTPases of the Rho and Ras family [9,25,26]. To determine if the effect of increased integrin activity observed in the Wehi 231 kd cells is caused by alterations in the activity of one of the factors mentioned, we inhibited Src kinases (PP1), PI3K (Wortmannin) and PLC (U73122) as central membrane proximal players by the use of pharmacological treatment. Figure 2C displays the highest inhibitor concentrations applied which still maintained cellular viability and were assumed to have the strongest effect on target activity. Wortmannin (100 mM) and U73122 (1 mM) treated co and kd cells showed a highly significant difference (WM co: 8.54%61.72%; kd: 27.52%61.95%; U73122 co: 5.91%61.31%; kd: 34.66%64.46%) in adhesion as observed for DMSO controls (co: 18.18%61.76%; kd: 47.72%62.15%), suggesting that although PLCc2 and PI3K contribute to overall adhesion, they are not causative for the observed increase in LFA-1 activity. Interfering with Src kinase activity (100 nM PP1) completely abolished adhesion in both co (1.21%60.39%) and kd (1.31%60.57%) cells, which likely positions the regulatory function of HPK1 downstream of Src.

HPK1 associates with the SKAP-HOM -RIAM complex and limits Rap1 activation
In T-cells, the ADAP-SKAP55 adaptor complex was shown to be of critical importance in T-cell adhesion by transducing signals from the TCR to integrins [15][16][17][18][19]. SKAP55 is known to interact with RIAM [18], an adaptor molecule previously shown to provide a physical link between the small GTPase Rap1 and talin [12] which binds to the b integrin chain as a central step in changing integrin affinity [4].
As it remains unclear whether an ADAP homologue in B-cells exists, we asked if HPK1 links to the SKAP55 homologue SKAP-HOM expressed in B-cells. Indeed, HPK1 co-immunoprecipitated with SKAP-HOM in Wehi 231 cells ( Figure 3A, upper two panels), providing a novel physical link to a B-cell adhesion regulatory module. In addition, as true for SKAP55 in T-cells [18], SKAP-HOM interacted with RIAM ( Figure 3A, third panel). Because HPK1 and RIAM both physically link to SKAP-HOM and the observed growth of the Wehi 231 kd cells resembled the reported behaviour of Rap1V12 transfected cell lines [27], we investigated Rap1-GTP expression in unstimulated and anti-IgM (anti m) stimulated transfectants. In unstimulated knockdown cells, the initial amount of Rap1-GTP was twice as high as in the control cells, whereas antigen receptor stimulation led to comparable Rap1-GTP amounts in knockdown and control clones ( Figure 3B), suggesting a negative regulatory role of HPK1 on B-cell adhesion by limiting Rap1 activation.

Lack of HPK1 affects actin polymerization and focal adhesion kinase activity
RIAM was shown to interact with Profilin and Ena/VASP proteins to regulate actin polymerization and cell spreading [28]. We hypothesized that the increased amount of active Rap1 found in the kd cells could affect the activity of the downstream adaptor RIAM and therefore change cellular F-actin levels. Wehi knockdowns stained for F-actin by the use of Phalloidin-FITC showed a higher basal F-actin level ( Figure 3C) than co cells, a phenotype reported for cells with increased RIAM activity [28], which possibly explains the enhanced spreading of kd cells. In addition, kd cells exhibited significantly less de novo actin polymerization upon anti-IgM F(ab')2 stimulation ( Figure 3D) which could be the result of uncoupling the actin regulatory SKAP-HOM -RIAM complex from coordinated signals through the BCR.
An additional downstream target of Rap1 activity is the focal adhesion kinase (FAK). FAK expression was shown for Wehi 231 and A20 cells as well as for activated B-cells, where the protein was suggested to colocalize with LFA-1 and VLA-4 in a manner critically dependent on actin dynamics to regulate cell spreading [29]. In addition, FAK was shown to associate with the Arp2/3 complex, therefore linking integrin signalling directly with actin polymerization [30]. The loss of HPK1 in Wehi 231 kd cells led to a constitutive and profound increase in the amount of active FAK phosphorylated on Y379 (pFAK379), while anti-IgM (anti m) stimulation had no influence on pFAK379 levels either in co or kd cells ( Figure 3E), which indicates an integrin-mediated FAK activation mechanism independent of the BCR. Thus, we provide evidence that through the lack of HPK1, enhanced Rap1 activity affects actin dynamics and the phosphorylation of the focal contact component FAK.

Murine B-cell development is largely unaffected by the lack of HPK1
Recently, HPK1 2/2 mice were reported to have normal numbers of B220 + cells and B-cell/T-cell ratios in peripheral lymphoid organs [31]. In line with these findings, our detailed analyses of B-cell subsets in the bone marrow showed that mice lacking HPK1 are only mildly affected and develop largely normal cell numbers. We observed a slight enlargement of the Hardy fraction C, composed of B220 + CD43 + BP1 + HSA int . pro-B cells ( Figure 4B) which are known to be in close contact with the stromal environment via CD44 and VLA-4 to ensure functional pro-pre-B-cell transition [32]. This leads to an increase in the small pre- (Figure 4C, Fr.D) and the mature B220 + CD43 2 B-cell population ( Figure 4A). Nevertheless, the slightly increased cellularity of the mature B-cell fraction in the bone marrow of HPK1 2/2 mice did not significantly affect the numbers of transitional (T1: B220 + CD93 + IgM + CD23 2/low , T2: B220 + CD93 + IgM + CD23 + , T3: B220 + CD93 + IgM low CD23 + ) and mature (B220 + CD93 2 )/marginal zone (B220 + CD1d + ) B-cells in the spleen ( Figure 5A,B). Interestingly, older HPK1 2/2 mice (17 w) exhibited increases in splenic leukocyte numbers compared to wt mice ( Figure 5C). Aged (40 w) HPK1 2/2 mice showed reductions of surface IgM density in transitional B-cells ( Figure 5D) which leads to a significant increase in the transitional 3/IgM low fraction ( Figure 5E), a phenotype that in an attenuated form reflects the B-cell development observable in anti-HELtg/HELtg mice [33]. The findings of Shui et al. [31] additionally stated a stronger overall increase in specific immunoglobulin levels after Tcell dependent protein antigen administration in HPK1 2/2 mice, which was suggested to be the result of cognate T-cell hyperreactivity. An alternative explanation of the described Bcell hyperreactivities could be a cell intrinsic defect in limiting Rap1-dependent integrin activation known to be of critical importance in refining antigenic activation thresholds.
Altered B-cell reactivity and BCR dynamics in HPK1 2/2 mice To assess B-cell reactivity in HPK1 2/2 mice naïve splenic Bcells were purified untouched, CFSE labelled and subjected to in vitro proliferation by the use of different B-cell stimuli. Anti-IgM (5 mg/ml)/anti-CD40 (10 mg/ml) as well as LPS (25 mg/ml) and LPS (25 mg/ml)/IL-4 (5 ng/ml) induced proliferation was significantly increased in HPK1 2/2 mice ( Figure 6A,C and Figure S3), while anti-CD40/IL-4 stimulation resulted in the weakest differences, suggesting a negative regulatory effect of HPK1 on BCR-and TLR-4 mediated B-cell activation operating independently on CD40 signalling ( Figure 6B and Figure S3). We further speculated whether the phenotype of altered actin dynamics, as shown for the Wehi knockdowns, as well as the reported association of HPK1 with the actin regulatory protein HS1 [34][35][36][37], might cause altered actin-mediated BCR uptake kinetics. Therefore we performed anti-IgM-FITC internalization assays with primary splenic cells. Splenic IgM + cells of HPK1 2/2 mice displayed an unsignificant tendency (n = 4) for a slower BCR endocytosis rate ( Figure S1A), an effect that became more obvious by knocking down HPK1 expression in CH27 cells ( Figure S1C). CH27 B-lymphoma cells display a transitional maturation phenotype (IgM + IgD + ) and, in contrast to Wehi 231 cells, efficiently internalize antigen receptors upon stimulation. HPK1 kd CH27 cells displayed significantly slower IgM uptake kinetics after 5 and 10 min on 37uC compared to the co target ( Figure  S1B), an effect gradually lost with increasing HPK1 concentration as CH27 clones with a less efficient knockdown showed internalization rates insignificantly different from co cells (data not shown). Thus, we show that while B-cell developmental processes are largely unaffected by the lack of HPK1, antigen receptor-and TLR-4-induced proliferation is limited by the action of HPK1.

Discussion
In this paper we provide evidence for a novel function of HPK1 in regulating B-lymphocyte adhesion. By associating with the SKAP-HOM -RIAM module, HPK1 limits Rap1-GTP-mediated signals aiming to upregulate integrin activity upon antigen receptor triggering (Figure 7). B-cells lacking SKAP-HOM were reported to be deficient in b1 and b2 integrin-mediated adhesion and exhibit a mild decrease in anti-IgM induced in vitro proliferation, while displaying normal membrane-proximal signalling upon BCR triggering and systemic immunoglobulin responses to T-dependent antigens [22]. HPK1 2/2 B-cells intriguingly oppose this phenotype, as in our hands they showed wild type-like specific serum Ig levels after administration of T-dependent antigen ( Figure S2) but enhanced in vitro proliferation upon anti-IgM/IL-4 and LPS stimulation. Interestingly, a similar immunization approach on the C57/BL6 background done by Shui et al. [31] showed significant differences in all Ig isotype levels investigated. Our hypothesis is that the negative regulatory function of HPK1 might vary with the genetic background and in our study is linked to a rather age-related effect. In this regard one might speculate whether a biased negative selection in the bone marrow due to a change in signal quality from the BCR could possibly lead to a larger pool of autoreactive peripheral T3/ IgM low B-cells as recently described in mice [33]. The caveat of the knockout model addressed is that the actions of HPK1 could be compensated on the cellular level by e.g. additionally affecting regulatory T-cell function. The involvement of HPK1 in various signalling pathways implies a complex pattern of regulation which might partially account for the controversial results gained from Tcell studies on HPK1-mediated regulation of AP-1 activity and IL-2 transcription [23,[38][39][40][41]. This necessitates the generation of cell specific and inducible knockout approaches to further delineate the exact nature of regulation in a systemic setting.
Although we found an upregulation of Rap1-GTP levels in cells lacking HPK1, the question of the guanosine nucleotide exchange factor responsible for the enhanced Rap1 activation in this setting is still unclear. A possible way to dampen Rap1 activity could be the reported association of HPK1 with CrkL [39], which was shown to bind the Rap1 exchange factor C3G [42] and localizes to the immunological synapse in a WAVE2 dependent manner [43]. Our results further indicate a function of HPK1 in the indirect control of cellular F-actin dynamics possibly by limiting the action of the downstream Rap1 effector RIAM.

Static B-cell adhesion, spreading and inhibitor treatment
Static adhesion assays were performed as described [13]; fluorescence of adhering cells is displayed as % remaining fluorescence after 5 consecutive washing steps [Fl rem. /Fl tot .6100]; for spreading assays, cells were preincubated with PP1 (100 mM), Wortmannin (100 nM), U73122 (1 mM) or DMSO for 30 min on  37uC and seeded onto ICAM-1/Fc layers for additional 30 min; phase contrast images were taken on a Nikon Eclipse microscope to determine the percentage of spread (phase dark) cells by counting four random fields.

Immunoprecipitation and Rap1-GTP pulldown
Cells (5610 6 /sample) were lysed in 1% Triton X-100, 10 mM Tris (pH 7.5), 150 mM NaCl, 5 mM EDTA, 10 mM NaF lysis buffer supplemented with protease (complete Mini; Roche) and phosphatase (Phosstop; Roche) inhibitors; lysates were precleared by centrifugation and subjected to immunoprecipitation for 3 h at 4uC; precipitates were washed, boiled in reducing loading buffer and separated on 4-12% gradient SDS-PAGE gels (BIORAD); Rap1-GTP pulldowns were performed with Ral GDS RBD agarose according to manufacturers instructions (Millipore); for quantification of Rap1-GTP amounts, co and kd band intensities were normalized for total Rap1 levels of lysates by densitometric evaluation (ImageJ) and all time points of Rap1-GTP samples were compared to the control target at t 0 = 1.

BCR internalization assay
BCR internalization was measured as previously described [45]. In brief, cells (1610 6 /sample) were stained on ice with anti-IgM-FITC (5 mg/ml), shifted on 37uC for various time points and reactions were stopped by adding an excess of ice cold PBS/3% FCS; Mean fluorescence intensities (MFIs) of IgM + cells were immediately analysed by flow cytometry and are displayed as % internalized receptor [(2MFI t /MFI t0 6100)+100].