Decoding the signaling profile of hematopoietic progenitor kinase 1 (HPK1) in innate immunity: A proteomic approach

Signaling via β2 integrins (CD11/CD18) as well as TCRs and BCRs involves similar pathways. However, the activation of the same signaling molecule can result in opposing effects. One such example is the hematopoietic progenitor kinase 1 (HPK1), which negatively regulates T and B cell activation but enforces neutrophil adhesion via β2 integrins. This difference may be defined by specific HPK1 interacting networks in different leukocyte subsets which have already been described in the adaptive immune system. Here, we set out to identify interacting proteins of HPK1 in neutrophil‐like differentiated HL‐60 cells exposed to immobilized fibrinogen and left nonactivated or Mn2+‐activated to allow β2 integrin‐dependent adhesion. Co‐IP experiments followed by mass spectrometry led to the identification of 115 HPK1‐interacting proteins. A total of 58 proteins were found only in nonactivated cells and 39 proteins only in Mn2+‐activated adherent cells. From these results, we decoded a pre‐existing signaling cluster of HPK1 in nonactivated cells encompassing proteins essential for β2 integrin‐mediated signaling during neutrophil trafficking, namely DNAX‐activation protein 12 (DAP12), spleen tyrosine kinase (Syk), and Rac1. Thus, our study provides novel insights into the complex architecture of the signaling processes during neutrophil activation and the complex signaling profile of HPK1 in leukocytes.


Introduction
The fine-tuned spatiotemporal response of leukocytes to external stimuli is a hallmark of the immune system. Sensing of stimuli occurs through a wide variety of different cell surface receptors, including integrins, which transmit signals into the cell. Signaling of β 2 integrins (CD11/CD18), as well as TCRs and BCRs, leukocyte protein of 76 kDa (SLP-76), adhesion, and degranulation promoting adapter protein (ADAP) and/or hematopoietic progenitor kinase 1 (HPK1) [3][4][5][6][7][8]. While most of these signaling molecules reinforce the initial signal along the cascade, leading to the execution of cell effector functions, such as antibody production, cell killing, and removal of pathogens, HPK1 has been reported, depending on the cell type, to positively or negatively regulate cell effector functions.
In general, HPK1 is regarded as a negative regulator of TCR and BCR signaling [6,9,10]. In T cells, HPK1 phosphorylates the adaptor proteins SLP-76 and growth factor receptor-bound protein 2 (Grb2)-related adaptor protein downstream of Shc (GADS), resulting in the interruption of TCR signaling [5,6,11]. Thus, HPK1 activity dampens T cell activation and T cell-mediated immune responses. Similarly, HPK1 phosphorylates the B cell linker protein (BLNK) upon BCR activation, inducing degradation of BLNK, thereby terminating BCR signaling [9]. Furthermore, β 2 integrin signaling in T and B cells is negatively regulated by HPK1 [8,12]. Here, HPK1 competes for the SLP-76 binding site with the adaptor protein complex ADAP-Src kinase-associated phosphoprotein of 55 kDa (SKAP55) that is critical for β 2 integrin activation [7,13]. This leads to attenuated activation of the β 2 integrin lymphocyte function-associated antigen 1 (LFA-1, CD11a/CD18) and, thus, impaired binding to its ligand ICAM-1 [8,12]. In sharp contrast, HPK1 is a positive regulator of β 2 integrin signaling in neutrophils [14]. Here, HPK1 mediates the induction of the high affinity state of the β 2 integrin LFA-1 upon stimulation with chemokine (C-X-C motif) ligand 1 (CXCL-1). This effect depends on inside-out activation of LFA-1 since bypassing inside-out signaling with Mn 2+ completely restores the ICAM-1 binding ability of HPK1-deficient neutrophils. Accordingly, HPK1-deficient neutrophils show reduced CXCL-1-mediated adhesion to immobilized ICAM-1 under flow conditions in vitro and neutrophil adhesion is impaired in postcapillary venules of the cremaster muscle upon stimulation with TNF-α in HPK1-deficient mice compared to WT mice [14].
The exact signaling pathways and/or binding partners of HPK1 in neutrophils are unknown to date, while the interacting proteins of HPK1 in lymphocytes have already been analyzed in detail [10,15]. As the function of HPK1 may be defined by its signaling network, we set out to decipher the interacting proteins of HPK1 in myeloid cells by co-IP using GFP Nano-Trap technology and subsequent mass spectrometry (MS).

Identification of the HPK1 interacting partners in dHL-60 cells
Next, dHL-60 HPK1-EGFP cells and dHL-60 EGFP control cells were used to identify HPK1-interacting proteins ( Fig. 2A). Co-IP experiments using the GFP Nano-Trap technology were performed with whole-cell lysates of dHL-60 HPK1-EGFP cells and dHL-60 EGFP control cells exposed to immobilized fibrinogen and either left nonactivated for control or activated with Mn 2+ to allow β 2 integrin-mediated adhesion. Subsequently, interacting proteins were subjected to MS analysis. The successful and highly specific IP of HPK1-EGFP in dHL-60 HPK1-EGFP cells but not in dHL-60 EGFP control cells was confirmed in Western blot studies (Fig. 2B). Here, the HPK1-EGFP fusion protein was pulled down in dHL-60 HPK1-EGFP cell lysate precipitates as well as endogenous HPK1 as HPK1 can form dimers [16]. As expected, no HPK1-EGFP fusion protein or HPK1 was detectable in dHL-60 EGFP control cell lysates.
Following MS, the identified peptides were further analyzed with MaxQuant software (Supporting Information Table S1) [17]. Here, proteins in dHL-60 HPK1-EGFP cells with a log 2 fold change >2 compared to dHL-60 EGFP control cells were defined as HPK1specific interactors. Eight nonspecific protein hits were excluded by CRAPome database analysis [18]. In total, 115 proteins interacting with HPK1-EGFP were found. Here, 58 proteins were only detected in nonactivated, suspended cells and 39 proteins were found only in Mn 2+ -activated, adherent cells (Fig. 2C, Supporting Information Table S2). Moreover, 18 proteins were coimmunoprecipitated with HPK1 in both conditions. HPK1 (gene MAP4K1) itself was detected in both samples as expected.

Interacting network of HPK1 in dHL-60 cells
To decipher the HPK1 signaling profile, we focused on proteins with a known function in signaling including transmembrane, membrane-associated, adaptor, cytoplasmic, and cytoskeletonassociated proteins (Table 1). Among the resulting 39 signaling proteins, 21 proteins were found in nonactivated cells only (Fig. 3A), 12 proteins were detected in Mn 2+ -activated cells only (Fig. 3B), and six proteins were found in both samples. From these 39 HPK1-interacting proteins, we constructed the interacting network of HPK1 by analyzing them for known interactions with each Endogenous HPK1, HPK1-EGFP, EGFP, and actin were detected by immunoblotting using specific antibodies. Representative image of three independent experiments with three biological replicates. (C) Confocal microscopic images of dHL-60 HPK1-EGFP cells expressing HPK1-EGFP (green) stimulated with fMLP exposed to immobilized fibrinogen and stained for mAbp1 (red) and F-actin (cyan) using a specific antibody and phalloidin, respectively. Scale bar, 10 μm. Representative image of three independent experiments with three biological replicates.
other and grouping them accordingly, using the STRING database with stringent parameter settings to only include protein-protein interactions with the highest confidence [19] (Fig. 3A and B). Next, we defined clusters of proteins that interact not only with HPK1 but also with an interactor of HPK1 in nonactivated and activated dHL-60 HPK1-EGFP cells. Interestingly, we found in nonactivated cells a network of 11 proteins (cluster 1) and a second, smaller network with four proteins (cluster 2) (Fig. 3A). In contrast, in activated cells, only two small clusters with known protein-protein interactions were identified, containing five and three proteins, respectively (Fig. 3B, clusters 3 and 4), suggesting that HPK1 is integrated into different signaling complexes in nonactivated and activated cells.
In summary, we decoded the signaling profile of HPK1 in myeloid cells and identified a pre-existing signaling cluster in nonactivated cells containing DAP12, Syk, and Rac1. However, upon induction of adhesion, HPK1 leaves this preformed signaling environment critically involved in rapid β 2 integrin-dependent

Discussion
In the present study, we identified the interacting proteins of HPK1 in nonactivated, suspended, as well as in Mn 2+ -activated, adherent dHL-60 HPK1-EGFP cells. Among the identified 115 proteins interacting with HPK1, we focused on those proteins with known functions in signaling and constructed the interacting network of HPK1. Within the interacting network, a subset of proteins formed a signaling cluster with HPK1 specifically in nonactivated cells, including the proteins DAP12 and Syk, known to be involved in β 2 integrin signaling in neutrophils [1,3,21,27]. Syk and the related tyrosine kinase zeta-chain-associated protein kinase 70 (ZAP-70) expressed in T cells are well-known interactors of HPK1 in leukocytes, inducing HPK1 activation [28][29][30]. However, in T cells, the activation of HPK1 leads to termination of TCR signaling, whereas in neutrophils, HPK1 is necessary to activate the β 2 integrin LFA-1, leading to neutrophil adhesion [6,14]. In nonactivated dHL-60 cells, Rac1 was identified as HPK1 interactor together with its activator, ARHGEF6. Rac1 is an important regulator of neutrophil adhesion by inducing cytoskeleton remodeling [22,31,32]. Interestingly, Rac1 activity has been shown to be downregulated upon β 2 integrin-dependent adhesion of human neutrophils [33]. The Rho GTPase dissociation inhibitor alpha (ARGHDIA, gene ARGHDIA), which can inhibit Rac1 activity, was found in activated, adherent cells. This suggests that HPK1 represents a functional switch which supports Rac1 activation in non-adherent cells by recruitment of ARHGEF6 and enforces Rac1 deactivation by dissociation from ARHGEF6 and association with ARGHDIA upon induction of adhesion. Association of HPK1 with DAP12 and Syk as well as Rac1 and ARHGEF6, therefore, suggests that HPK1 composes or is at least part of a preformed protein complex necessary for rapid induction of β 2 integrin signaling in adherent neutrophils. A considerable difference between the roles of HPK1 in neutrophil recruitment and T cell activation may be due to the different time course and context of these two processes. Free-flowing neutrophils have to be prepared for the rapid induction of β 2 integrin-dependent adhesion and subsequent arrest on inflamed endothelial cells. Therefore, the proteins necessary for β 2 integrin signaling may exist in a preformed signaling complex in which HPK1 may play a central role as an adaptor protein, providing the molecular machinery for rapid neutrophil activation while the cell is still inactive [34]. In this context, HPK1 may act in analogy to lymphocytes as a negative regulator which keeps its interacting partners in an inactive state until the induction of adhesion. Thus, HPK1 may be important for sustained signaling processes, promoting an inactive state. This is in line with the finding that the absence of HPK1 promotes hyperinflammation [6].
To date, it is unclear whether the kinase activity of HPK1 is necessary for β 2 integrin signaling in neutrophils, as in previous studies neutrophils were used from HPK1 KO mice [14]. Since HPK1 is harboring proline-rich motifs as well as a citron homology domain enabling protein-protein interactions, HPK1 could at least partly exert its functions as scaffold molecule [35,36]. The use of HPK1 kinase-dead mouse strains showed that dampening TCR signaling by HPK1-induced destabilization of SLP-76 complexes depends on HPK1 kinase activity in T cells [36]. In contrast, HPK1 interactions with Grb2, GADS, and mAbp1, among others, apparently do not involve phosphorylation of these adaptors by HPK1 [29,30,[36][37][38][39]. Similarly, a scaffold function has also been shown for the kinase Syk linking the B cell-specific adaptor BLNK to Ca 2+ mobilization [40]. Accordingly, HPK1 may recruit and/or stabilize a protein complex including all components necessary for rapid β 2 integrin signaling and neutrophil arrest independent from its kinase activity [36]. However, substrates of HPK1 will probably not be identified by co-IP and subsequent MS as the interaction between kinases and their substrates is often not stable enough to survive this procedure. Here, phosphoproteomic or HPK1 kinasedead co-IP approaches would be helpful to distinguish signaling processes dependent on HPK1 kinase function and/or a proposed HPK1 scaffold function.
In conclusion, our data indicate that HPK1 may function as a potential scaffold protein providing a pre-existing protein cluster for rapid β 2 integrin signaling and neutrophil postarrest functions. Surprisingly, the majority of the identified interacting proteins of HPK1 in neutrophils were not previously described as HPK1 interactors in T and B cells which may explain the different functions of HPK1 in neutrophils versus T and B cells [10,15]. Some of the proteins found to interact with HPK1 in neutrophils, such as CD157, platelet endothelial cell adhesion molecule 1 (PECAM-1), uPAR, and Arf6, have already been reported to regulate neutrophil adhesion, spreading, chemotactic migration, and extravasation as well as postadhesion functions [23,25,26,41]. Recently, HPK1 has emerged as potent target for cancer therapy [42,43]. Here, suppression of HPK1 function enhances antitumorigenic behavior of T cells and DCs and HPK1 inhibitors specifically targeting HPK1 kinase activity are currently being designed as therapeutic molecules. In this context, the effect of HPK1 inhibition on neutrophil function has to be carefully considered.

Generation of HL-60 HPK1-EGFP cells
The coding sequence of human HPK1 (Source BioScience, United Kingdom) was cloned into a pEGFP-N1 vector to generate the HPK1-EGFP fusion sequence which was subcloned into the FuVal lentiviral vector pFu-P2A-Puro using the In-Fusion cloning kit (Takara Bio Europe, France). The resulting pFu-HPK1-EGFP-P2A-Puro vector was used to transfect HEK-293T/1744]. Virus-containing supernatant was harvested 48 h post-transfection and used to transduce HL-60 cells, which were subsequently cultured in the presence of 1 μg/mL puromycin (Sigma-Aldrich, Germany) for 3 weeks. Cells were sorted according to EGFP fluorescence. Expression of the HPK1-EGFP fusion protein was detected by Western blotting.
Confocal microscopy was performed at the Core Facility Bioimaging of the Biomedical Center of the Ludwig-Maximilians-Universität München using an upright confocal Leica SP8X WLL microscope (Leica, Germany).

Co-immunoprecipitation
Co-IP experiments were conducted with GFP Nano-Trap beads (Chromotek, Germany) and performed as described before [44]. In brief, dHL-60 HPK1-EGFP or dHL-60 EGFP control cells were exposed to immobilized fibrinogen (250 μg/mL, Sigma-Aldrich) for 15 min and either left nonactivated as control or activated with 1 mM Mn 2+ to allow β 2 integrin-mediated adhesion. Successful performance of the co-IP was analyzed via Western blot technique and the immunoprecipitates were subjected to MS.

Mass spectrometry
GFP Nano-Trap beads with bound proteins were washed with 50 mM NH 4 HCO 3 and digested with 10 ng/μL trypsin in 1 M urea and 50 mM NH 4 HCO 3 . The overnight digestion was performed in the presence of 1 mM DTT. Digested peptides were alkylated and desalted before LC. For LC-MS purposes, desalted peptides were injected in an Ultimate 3000 RSLCnano system (Thermo Fisher Scientific), separated in a 15 cm analytical column (75 μm ID with ReproSil-Pur C18-AQ 2.4 μm, Dr. Maisch GmbH, Germany) with a 50 min gradient from 5 to 60% acetonitrile in 0.1% formic acid. The effluent from the HPLC was directly electrosprayed into a QExactive HF (Thermo Fisher Scientific) operated in a data-dependent mode to automatically switch between full-scan MS and MS/MS acquisition. Survey full-scan MS spectra (from m/z 375 to 1600) were acquired with resolution R = 60,000 at m/z 400 (AGC target of 3 × 10 6 ). The ten most intense peptide ions with charge states between 2 and 5 were sequentially isolated to a target value of 1 × 10 5 , and fragmented at 27% nor-malized collision energy. Typical mass spectrometric conditions were: spray voltage, 1.5 kV; no sheath and auxiliary gas flow; heated capillary temperature, 250°C; ion selection threshold, 33.000 counts.
MaxQuant version 1.5.2.8 software was used for protein identification and quantification using the intensity-based absolute quantification (iBAQ) value [17]. Proteins were considered as relevant interaction partners of HPK1 if the difference of the iBAQ value (log 2 fold-change) between HPK1-EGFP and EGFP co-IP experiments was higher than 2. To identify and eliminate nonspecific interacting proteins, a CRAPome analysis was conducted as described previously [18]. The MS proteomics data have been deposited to the ProteomeXchange Consortium via the PRIDE partner repository with the dataset identifier PXD021136 [45].

Generation of the HPK1 interacting network
The interacting network of HPK1 was analyzed and visualized using Cytoscape 3.8.0 software [46]. All interacting proteins of HPK1 identified in our screen were manually categorized into transmembrane, membrane-associated, adaptor, cytoplasmic, and cytoskeleton-associated proteins based on gene ontology terms according to The Human Protein Atlas and those with known functions in signaling were further considered [47]. These HPK1-interacting proteins were merged with known interactions extracted from the STRING database [19]. The shown interactions were limited to those found in experiments or databases, and only those interactions with highest confidence according to STRING parameters (score > 0.9) were considered.