Thromb Haemost 2022; 122(11): 1943-1947
DOI: 10.1055/s-0042-1751280
Letter to the Editor

Bacterial-Type Long-Chain Polyphosphates Bind Human Proteins in the Phosphatidylinositol Signaling Pathway

Viola Krenzlin*
1   Center for Thrombosis and Hemostasis, University Medical Center Mainz, Mainz, Germany
,
Julian Roewe*
1   Center for Thrombosis and Hemostasis, University Medical Center Mainz, Mainz, Germany
,
Marcel Strueve*
1   Center for Thrombosis and Hemostasis, University Medical Center Mainz, Mainz, Germany
,
María Martínez-Negro
2   Department of Physical Chemistry of Polymers, Max Planck Institute for Polymer Research, Mainz, Germany
,
Arjun Sharma
1   Center for Thrombosis and Hemostasis, University Medical Center Mainz, Mainz, Germany
3   Department of Medicine, Pulmonary Center, Boston University School of Medicine, Boston, Massachusetts, United States
,
1   Center for Thrombosis and Hemostasis, University Medical Center Mainz, Mainz, Germany
,
Svenja Morsbach
2   Department of Physical Chemistry of Polymers, Max Planck Institute for Polymer Research, Mainz, Germany
,
1   Center for Thrombosis and Hemostasis, University Medical Center Mainz, Mainz, Germany
3   Department of Medicine, Pulmonary Center, Boston University School of Medicine, Boston, Massachusetts, United States
› Author Affiliations
Funding This work was supported by financial resources obtained from the Federal Ministry of Education and Research (01EO1503 to M.B.), the Deutsche Forschungsgemeinschaft (BO3482/3–3, BO3482/4–1 to M.B.), and the National Institutes of Health (R01AI153613 to M.B.). C.R. was awarded a Fellowship of the Gutenberg Research College at the Johannes Gutenberg-University Mainz.
 

Introduction

Zoom Image

Inorganic polyphosphates are linear polymers of monophosphate residues (Pi) that exist as short chains (Pi30–120) in platelet δ-granules and as long chains in bacteria.[1] A higher chain length increases the activity of these anionic polymers to accelerate factor XIIa-mediated factor XI activation, thrombin generation, block tissue factor pathway inhibitor activity, and strengthen fibrin clots by enhancing their mechanical stability and resistance to fibrinolysis.[2] [3] In bacteria, polyphosphates are associated with energy and phosphate storage, stress resistance, chelation of metal ions, and escaping host immunity.[4] During severe infections, long-chain polyphosphates from bacteria contribute to coagulopathy, neutrophil extracellular trap formation, and vascular-endothelial dysfunction.[3] [5] [6] However, the mechanisms that convey the pleiotropic activities of polyphosphates in living cells remain understudied.


#

Results and Discussion

Here, we report a screen using a human proteome microarray to identify novel polyphosphate-binding proteins. This HuProtTM microarray contains a large number of eukaryote-expressed proteins individually printed on a single array slide. A total of 19,394 unique proteins representing approximately 75% of the human proteome were evaluated. In total, 14,906 proteins could be assigned to cell physiological processes ([Fig. 1A]). The microarray slides were probed with biotinylated long-chain polyphosphates followed by fluorescence detection ([Fig. 1B]; full dataset: https://doi.org/10.5281/zenodo.5748254).

Zoom Image
Fig. 1 A human proteome microarray for the identification of polyphosphate-binding proteins. (A) Overview of the proteins in the HuProtTM human proteome microarray. In total 19,394 human proteins were assessed and 14,906 (multiple mapping possible) were assigned to cell physiological processes using the PANTHER software. (B) Representative microarray slides probed with polyphosphates and controls showing the best hit DAB1. The left array shows the subarray with polyphosphates and the right panel depicts a control array. The bottom rows of the subarrays contain control spots; box 1: histones; boxes 2 and 4: biotinylated bovine serum albumin; boxes 3 and 5: directly fluorophore-labeled IgG. No specific protein interactions were observed for the subarray on the control slides (right panel). (C) Mean fluorescence intensities of polyphosphate-binding signals for the top 30 proteins in the microarray. The arithmetic mean of technical duplicates is shown and the horizontal dashed line indicates the assigned threshold. (D) Isothermal titration calorimetry of long-chain polyphosphates titrated to either recombinant human DAB1 or human serum albumin (HSA). (E) Concentrations of selected proteins in human plasma. The data were retrieved from the open-access Peptide Atlas (plasma non-glyco 2017) acquired through mass spectrometry-based proteomics. (F) Dot plot showing mRNA expression of DAB1 and PIP4K2B in human tissues and selected cell types. Data were retrieved from the Human Protein Atlas. (G) Images of fixed, permeabilized HeLa cells stained with antibodies for DAB1, biotinylated polyphosphates, merged image, and DAPI for nuclei. (H) PIP4K2B and polyphosphate staining of HeLa cells; scale bars: 25 μm.

The best binding partner for polyphosphates was the protein DAB1 (Disabled-1; [Fig. 1C]). While DAB1 is a key player of the Reelin–DAB1 signaling pathway, regulating mammalian brain development by orchestrating cell migration, additional roles in nonneuronal tissues have become more evident in recent years.[7] Intracellular phosphorylation of DAB1 by Src family tyrosine kinases (Fyn/Src) facilitates downstream effector recruitment like PI3Ks (phosphatidylinositol 3-kinases), resulting in activation of multiple cellular signaling cascades. DAB1 interacts with PI3Ks in response to Reelin signaling.[8] DAB1 is phosphorylated after binding of activated protein C and PAR (protease-activated receptor) activation upstream of PI3Ks.[9] Reelin is also expressed by platelets and it was described that Reelin–platelet interactions can enhance platelet spreading on fibrinogen.[7] [10]

We confirmed the interaction of polyphosphates with human DAB1 using isothermal titration calorimetry (ITC) at 25°C. As shown in [Fig. 1D], polyphosphates interacted with recombinant DAB1 in an exothermic process with a heat release in the order of magnitude, which is usually observed for protein–protein interactions. However, it was not possible to fit the obtained data due to the low number of titration measurements performed with the limited amount of protein available. Since DAB1 does not contain a PASK domain for covalent attachment of polyphosphates,[11] [12] we speculate that the binding could be due to electrostatic interactions. We used titration of polyphosphates to human serum albumin (HSA) as a negative control without signs of binding as no significant heat evolvement was detectable ([Fig. 1D]). DAB1 is an intracellular protein and not detectable in human plasma ([Fig. 1E]). DAB1 gene expression was enriched in the brain among other tissues ([Fig. 1F]).[13] Functionally, DAB1 signaling in the endothelial layer of the brain orchestrates neuro-glia-vessel communication.[14]

The proteome microarray screen also indicated the binding of polyphosphates to PIP4K2B (phosphatidylinositol-5-phosphate 4-kinase 2B; [Fig. 1C]). PIP4K2B is ubiquitously expressed in many tissues including endothelial cells ([Fig. 1F]). PIP4K2B is a stress-regulated lipid kinase that catalyzes the phosphorylation of phosphatidylinositol-5-phosphate to form phosphatidylinositol-5,4-bisphosphate (PIP2), which is a substrate for the phosphatidylinositol signal transduction pathway. Loss of PIP4K2B is associated with slower tumor growth in p53-deficient mice.[15] [16]

Fluorescence microscopy revealed co-localization of antibody staining for DAB1 and PIP4K2B in HeLa cells probed with polyphosphates ([Fig. 1G, H]).

Several other binding proteins were observed in the microarray with detectable affinity, albeit with lower scores than the threshold recommended by the manufacturer:

MARCKSL1 (myristoylated alanine-rich C-kinase substrate-like 1, [Fig. 1C]) can sequester PIP2 within “lipid rafts” of the cell membrane for participation in later signal transduction events.[17] MARCKSL1 plays a critical role in the regulation of apoptosis and promotes antiangiogenic effects by reducing VEGF (vascular endothelial growth factor) and HIF-1α (hypoxia-inducible factor 1-alpha).[18]

PLEK (pleckstrin) is highly expressed in blood cells, especially platelets, bone marrow, and lymphoid tissue, and present in blood plasma ([Fig. 1C, E]).[13] In platelets, PLEK is a major substrate of protein kinase C. Upon binding to PIP2 it is involved in the regulation of multiple platelet processes, e.g., aggregation, degranulation, platelet activation, and secretion.[19] [20]

The NUDT3 (nucleoside diphosphate-linked moiety X-type motif hydrolase-3, [Fig. 1C]) gene encodes for the diphosphoinositol polyphosphate phosphohydrolase-1. It belongs to the MutT, or Nudix, protein family and cleaves a β-phosphate from the diphosphate groups in diphosphoinositol pentakisphosphate and bisdiphosphoinositol tetrakisphosphate, suggesting that it may play a role in signal transduction.[21] NUDT3 was identified in a gain-of-function screen as a PI3K inhibitor resistance gene suggesting a role in this pathway.[22] Interestingly, NUDT3 was recently reported to show polyphosphatase activity with a requirement of Zn2+ as a cofactor.[23]

NAMPT/visfatin (nicotinamide phosphoribosyltransferase) induces endothelial angiogenesis through activation of the PI3K pathway.[24] Among the top 10 polyphosphate-binding proteins, VCX3A (variable charge, X-linked 3A; unknown function), AKAP8L (A-kinase anchoring protein 8-like), GRHPR (glyoxylate reductase/hydroxypyruvate reductase; metabolism), and GPD1 (glycerol-3-phosphate dehydrogenase; metabolism; [Fig. 1C]) have no documented connection to the phosphatidylinositol pathway. NAMPT, GRHPR, GPD1, and NACA (nascent-polypeptide-associated complex alpha polypeptide), but none of the other proteins shown in [Fig. 1C], were also identified by another HuProtTM microarray screen probed with ultra-short polyphosphate (Pi8).[25] Likewise, GPD1 from Trypanosoma spp. is a binding partner of long-chain polyphosphates.[26]

In summary, the majority of the 10 proteins with highest polyphosphate binding were associated with the PI3K pathway. Interestingly, polyphosphates directly modulate PI3K/Akt signaling including in primary human vascular endothelial cells.[5] [27] The siRNA (small-interfering RNA) knockdown of PI3K/Akt/mTOR pathway components abrogates the endothelial barrier-disruptive effects of polyphosphates.[5] Short chains (Pi15) are sufficient to activate mTOR.[27] Hence, our studies expand the previous knowledge by uncovering several new polyphosphate-binding proteins with a relation to the PI3K/Akt pathway.

In contrast, our microarray screen could not confirm binding of polyphosphates to their putative receptors, RAGE and P2Y1,[28] [29] which is in line with their nonsignificant role for mediating polyphosphate effects in macrophages.[4] However, a negative result in the microarray does not exclude that a protein binds polyphosphates under physiologic conditions, as binding characteristics are affected by protein folding and posttranslational modifications. Notably, anionic polyphosphates partner with cations (e.g., Na+, K+, Ca2+, Mg2+, Zn2+), which affects the overall charge and binding characteristics of polyphosphates to proteins. Our current study did not investigate the influence of different cations on polyphosphate–protein interactions; or if short chain, platelet-derived polyphosphates can also bind the identified factors. Instead, we provide a comprehensive list of long-chain polyphosphate-binding human proteins to facilitate mechanistic studies on the molecular targets of polyphosphates in future.


#

Methods

Reagents: Long-chain polyphosphates (Pi700) were a gift from James H. Morrissey and Stephanie A. Smith (see Kerafast #EUI006 for further specifications). Human recombinant DAB1 was produced in Escherichia coli by Aviva Systems Biology (San Diego, California, United States).

Human Proteome Microarray: The HuProtTM v3 microarray was performed by Cambridge Protein Arrays Ltd., Babraham Research Campus, Cambridge CB22 3 AT, United Kingdom. Slides were probed with biotinylated long-chain polyphosphates (Pi700; 500 μM in 2% BSA/PBS [pH 7.45]-Tween 0.1%) followed by fluorescence detection using streptavidin-532.

ITC: The calorimetric measurements were performed using a NanoITC Low Volume (TA Instruments, Eschborn, Germany) with an effective cell volume of 170 µL. Briefly, 50 µL of a long-chain polyphosphate solution (20.8 µM in OPCA or PBS buffer) was titrated to a suspension of DAB1 (4 µM in OPCA buffer) or HSA (3 µM in PBS, negative control). The experimental temperature was kept constant at 25°C. Additionally, the same amount of polyphosphate solution was titrated into pure OPCA and PBS buffers to determine the heat of dilution for reference. The number and injected volume of the titration steps were the same for all measurements (25 × 2 µL). The spacing between injections was set to 300 seconds. The integrated reference heats were then subtracted from the integrated heats of the adsorption experiments. Due to the high costs and low availability of the DAB1 protein, the results are shown for one single series of measurements only. Thus, no fitting was performed. For calculating molar ratios, a length of 700 phosphate residues per chain was applied. Nano Analyze Data Analysis software from TA Instruments was used for the data evaluation of the ITC measurements (Software version 2.5.0) from TA Instruments.

Immunofluorescence microscopy: HeLa cells were fixed in 4% paraformaldehyde for 15 minutes, washed twice with PBS, incubated for 10 minutes with 0.25% Triton X-100, washed three times with PBS, blocked with 1% BSA in PBST for 30 minutes, and incubated with primary antibodies (1:100 dilution, rabbit anti-human DAB1, rabbit anti-human PIP4K2B; MyBioSource, San Diego, California, United States) and biotinylated polyphosphates (100 μM) at 4°C overnight. Next, slides were washed with PBS three times, incubated with secondary antibodies (1:200 dilution, goat anti-rabbit IgG-AF594; MyBioSource) and streptavidin-AF488 (ThermoFisher Scientific, Waltham, Massachusetts, United States) at room temperature in the dark for 1 hour, washed three times, and mounted with DAPI. Secondary antibody-only incubations served as negative controls and showed no unspecific staining. Images were acquired with a Nikon deconvolution wide-field epifluorescence system.

Bioinformatics: The consensus-normalized expressions in [Fig. 1F] were retrieved from the Human Protein Atlas (HPA) and are based on combined bulk transcriptome data from the HPA, Genotype-Tissue Expression (GTEx), and Functional Annotation of Mammalian Genomes 5 (FANTOM5) projects. Data for endothelial cells were retrieved from HPA from single cell transcriptomes. The colors represent the natural log (loge)-normalized RNA expression values plus one.[13]


#
#

Conflict of Interest

M.B. is a consultant for and receives research support from ARCA Biopharma.

Acknowledgments

We thank James H. Morrissey and Stephanie A. Smith for providing reagents and discussions. We thank Jena B. Goodman, Michael T. Kirber, Shivraj Yabaji, and the BUMC Cellular Imaging Core for assistance with fluorescence microscopy. We thank Archana Jayaraman and Johannes Platten for assistance with manuscript preparation and Lindsey Stein for secretarial assistance. C.R. is a scientist at the German Center for Cardiovascular Research (DZHK). The authors are responsible for the content of this publication.

* These authors contributed equally.



Address for correspondence

Markus Bosmann, MD
Department of Medicine
Pulmonary Center, Boston University School of Medicine
Boston, MA 02118
United States   

Publication History

Received: 10 December 2021

Accepted: 17 May 2022

Article published online:
30 July 2022

© 2022. Thieme. All rights reserved.

Georg Thieme Verlag KG
Rüdigerstraße 14, 70469 Stuttgart, Germany


Zoom Image
Zoom Image
Fig. 1 A human proteome microarray for the identification of polyphosphate-binding proteins. (A) Overview of the proteins in the HuProtTM human proteome microarray. In total 19,394 human proteins were assessed and 14,906 (multiple mapping possible) were assigned to cell physiological processes using the PANTHER software. (B) Representative microarray slides probed with polyphosphates and controls showing the best hit DAB1. The left array shows the subarray with polyphosphates and the right panel depicts a control array. The bottom rows of the subarrays contain control spots; box 1: histones; boxes 2 and 4: biotinylated bovine serum albumin; boxes 3 and 5: directly fluorophore-labeled IgG. No specific protein interactions were observed for the subarray on the control slides (right panel). (C) Mean fluorescence intensities of polyphosphate-binding signals for the top 30 proteins in the microarray. The arithmetic mean of technical duplicates is shown and the horizontal dashed line indicates the assigned threshold. (D) Isothermal titration calorimetry of long-chain polyphosphates titrated to either recombinant human DAB1 or human serum albumin (HSA). (E) Concentrations of selected proteins in human plasma. The data were retrieved from the open-access Peptide Atlas (plasma non-glyco 2017) acquired through mass spectrometry-based proteomics. (F) Dot plot showing mRNA expression of DAB1 and PIP4K2B in human tissues and selected cell types. Data were retrieved from the Human Protein Atlas. (G) Images of fixed, permeabilized HeLa cells stained with antibodies for DAB1, biotinylated polyphosphates, merged image, and DAPI for nuclei. (H) PIP4K2B and polyphosphate staining of HeLa cells; scale bars: 25 μm.