Y-box protein-1 is actively secreted through a non-classical pathway and acts as an extracellular mitogen

Monocytic cells secrete YB-1 on LPS incubation

In an experimental model of acute kidney-transplant rejection (Kunter et al, 2003), ED1-positive monocytes and macrophages showed a high level of YB-1 expression (Fig 1A). Given the missing confinement of YB-1 to cell boundaries in mesangioproliferative disease (van Roeyen et al, 2005) and its strong expression by mononuclear cells, we suggested that YB-1 secretion might have occurred. Secretion of a nuclear protein seems unexpected at first; however, such a process is not unprecedented, given the secretion of transcription factor HMGB1 in inflammation, a protein with similar evolutionary conservation.

Primary human monocytes were exposed to increasing lipopolysaccharide (LPS) concentrations for 4 h and the supernatants were trichloroacetic acid (TCA)-precipitated. Immunoblotting showed dose-dependent secretion of YB-1, which peaked at a concentration of 5 ng/ml LPS (Fig 1B), whereas no non-specific cytosolic protein release, for example that of glyceraldehyde 3-phosphate dehydrogenase, was observed. A Trypan-blue exclusion assay confirmed that protein release was not due to a lack of cell integrity (viability >95%). By contrast, intracellular YB-1 protein levels essentially did not change after LPS exposure (Fig 2B, lower panel). Calculation of protein loading and band intensities indicated that approximately 25% of total YB-1 protein was secreted by primary monocytes on stimulation, using 5 ng/ml LPS (Fig 1C). Quantification of macrophage migration inhibitory factor (MIF; Kleemann et al, 2000) in the same supernatant by ELISA indicated a similar dose response to LPS (Fig 1D). LPS-induced secretion of both YB-1 and MIF was also observed in MonoMac-6 (MM6) cells (supplementary Fig 1B–D online), underpinning the above findings. MM6 cells are well-characterized LPS-responsive monocytic cells. In this system, YB-1 secretion was also apparent in a narrow LPS concentration range between 1 and 7.5 ng/ml, which paralleled MIF secretion (supplementary Fig 1B,C online). Kinetic studies carried out at the peak LPS concentration of 5 ng/ml showed low-level baseline secretion of YB-1 up to 2 h, followed by an apparent marked LPS effect after 4 h (supplementary Fig 1D online).

On the basis of earlier observations made for HMGB1 (Gardella et al, 2002), we addressed the possibility whether vesicles are involved in YB-1 release. Vesicles were prepared from supernatants on cell stimulation with LPS (supplementary Fig 2A online). Indeed, vesicle-enriched fractions contained YB-1 (Fig 1E), and intravesicular YB-1 was protected from protease digestion (trypsin), unless vesicles were solubilized by detergent (Triton X-100) preincubation (Fig 1E). Contamination with other cytosolic proteins was excluded, given the negative glyceraldehyde 3-phosphate dehydrogenase immunoblot. Furthermore, successful vesicle enrichment was confirmed by anti-RAB7 western blot (Fig 1E). After the removal of vesicles by centrifugation, supernatants were analysed for YB-1 protein content. In fact, TCA precipitation showed the presence of YB-1 in these supernatants, indicating that after secretion vesicles disintegrate and YB-1 becomes freely available in the extracellular milieu (Fig 1E, sixth panel). Thus, monocytes stimulated by LPS release YB-1 within vesicles. Trypsin-digest experiments ruled out that YB-1 might have adhered non-specifically to the cytosolic side of the vesicular membrane. Importantly, YB-1 secretion, stimulated in a similar manner, was also observed in rat mesangial cells (rMCs). Mesangial cells share phenotypic and functional characteristics with monocytes (Mené et al, 1989) and have been shown to upregulate YB-1 under inflammatory conditions (van Roeyen et al, 2005). Consequently, we observed that the inflammation-related stimuli transforming growth factor-β and hydrogen peroxide also resulted in vesicular secretion of haemagglutinin-tagged YB-1 stably expressed in rMCs in a time- and dose-dependent manner (supplementary Fig 2B–D online). As, in this cell model, YB-1 was specifically detected through its haemagglutinin tag, any cross-reactivity with other cold-shock proteins could be excluded. Next, endogenous YB-1 release from rMCs was tested. Both LPS and platelet-derived growth factor-BB provoked a substantial release of YB-1 over a time range of 1–4 h (supplementary Fig 1A online).

YB-1 is secreted by an alternative secretion mode

Using confocal laser-scanning microscopy, intracellular vesicle formation with enriched levels of fluorescent YB-1–GFP (green fluorescent protein) fusion protein was detected within 2 h, but not in control cells expressing GFP alone (Fig 2A; see also live cell imaging in supplementary video online). These data hint at a non-classical secretion mode for YB-1, bypassing the endoplasmic reticulum and Golgi apparatus, as first described by Rubartelli et al (1990) for interleukin-1β (IL-1β). For several inflammatory mediators (IL-1β, MIF, HMGB1, thioredoxin 1 and fibroblast growth factor 2), non-classical secretion has now been verified (reviewed by Nickel, 2003). All these proteins lack an amino-terminal signal-peptide sequence, which is required for endoplasmic reticulum/Golgi targeting. In fact, computer-based algorithms predict non-classical secretion of YB-1 and indicate the absence of a canonical N-terminal signal peptide motif. To investigate further the secretion mode of YB-1, we applied several biochemical approaches. We first noticed that preincubation with brefeldin A, which interferes with the classical secretion pathway, did not reduce, but instead, increased, the secretion of YB-1 (Fig 2B), as shown earlier for other leaderless proteins such as MIF and IL-1β (Rubartelli et al, 1990; Flieger et al, 2003). For non-classically secreted proteins, different export machineries have been described. For example, IL-1β and MIF secretion depends on ABC transporters, as indicated by an inhibition of the secretion of these mediators by glyburide or probenicid (Andrei et al, 1999; Flieger et al, 2003). We tested whether glyburide and probenicid would interfere with the secretion of YB-1. MM6 cells were incubated with these agents before LPS stimulation. Both glyburide and probenicid treatment resulted in a marked reduction of YB-1 secretion (Fig 2C,D). Protein exclusion of leaderless proteins has been described to occur through vesicle formation; for example, through exosomes, endolysosomes or microvesicle shedding (Andrei et al, 1999; Gardella et al, 2002). To confirm this question with respect to YB-1 secretion, we pursued two pharmacological approaches: one using ionomycin, which is a potent ionophore that induces degranulation in some cell types, and the other using reserpine, which can disrupt secretion by acting as an inhibitor of the ATP-dependent uptake of bioamines into vesicles. Ionomycin was found to have a stimulatory effect on the LPS-induced secretion of YB-1 (Fig 2E), whereas preincubation of MM6 cells with reserpine strongly reduced YB-1 secretion (Fig 2F), mirroring observations made earlier for IL-1β using similar drugs (Andrei et al, 1999). It should be noted that non-specific YB-1 release due to cell death was excluded in all experiments, as indicated by the lack of measurable lactate dehydrogenase release. Together, these results are in line with vesicle-mediated YB-1 secretion. We next coexpressed GFP–RAB7 and dsRed–YB-1 in mesangial cells. Whereas in non-stimulated cells dsRed–YB-1 is diffusely distributed throughout the cytosolic compartment, LPS stimulation led to intravesicular accumulation of YB-1 and to a notably partial, yet substantial, colocalization with GFP–RAB7 (supplementary Fig 3 online). This indicated that YB-1 secretion occurred through the endolysosomal compartment.

Acetylation of HMGB1 is an essential step for protein redistribution and its secretion (Bonaldi et al, 2003). YB-1 contains 16 lysine residues (Fig 2G). Two lysines at positions 301 and 304 at the carboxy-terminus of the protein are subject to acetylation (data not shown). Exchange of these lysines by alanines by using site-directed mutagenesis fully abrogates the secretion of Flag-tagged YB-1, which is otherwise seen at 2 h after LPS addition (Fig 2H).

In summary, these results unanimously show that YB-1 is secreted through an alternative secretion mode similar to that observed for other leaderless proteins such as HMGB1, IL-1β or MIF.

Extracellular YB-1 stimulates migration and proliferation

Extracellular occurrence of YB-1 raises the question of its functional relevance. YB-1 orchestrates proliferation, for example through DNA polymerase-α expression (Kohno et al, 2003; En-Nia et al, 2005). Therefore, we studied the proliferative response of different cell lines on stimulation with extracellular YB-1. rMCs as well as human kidney 2 (HK-2) cells were assayed for their DNA synthesis rates by using 5-bromodeoxyuridine incorporation. Synchronized serum-depleted cells were stimulated with purified rYB-1 (0.1 μg/ml and 1 μg/ml) or FCS (10% vol/vol, as a positive control). Addition of rYB-1 increased the DNA synthesis rate of both cell lines in a dose-dependent manner (supplementary Fig 3A,B online). rYB-1 at 1 μg/ml had a stimulatory effect of 90% (HK-2) and 40% (rMC), and the results were 35% and 60%, respectively, with FCS. Specificity of the pro-proliferative effect of rYB-1 was underscored by blockade of the YB-1 effect with a monoclonal YB-1 antibody (van Roeyen et al, 2005). Similar results were obtained with breast epithelial cells of cancerous (BT20) and non-cancerous (MCF12A) origin (data not shown). Addition of YB-1 for 24 h increased the proliferation rate of both cell lines to a similar degree as that achieved by the addition of epidermal growth factor (EGF; positive control). Denatured YB-1 had only a marginal influence on proliferation, ruling out any effect because of endotoxin contamination of the YB-1 preparation.

Besides mitotic events, cell motility is an important function that relates to inflammatory processes with a turnover of matrix and alterations of tissue architecture. Changes in proliferation as well as motility were assessed in scratch-wound assays, in which the cell monolayer is ‘injured'. Closure of this scratch was significantly and dose-dependently inhibited by the addition of monoclonal YB-1 antibody (supplementary Fig 3C online) compared with isotype-matched control antibody, showing that wound closure was dependent on YB-1.

In a reciprocal approach, mesangial cells were grown to confluency and a scratch was introduced using a rubber policeman. Mesangial cell migration was monitored by light microscopy for 48 h under non-stimulated conditions and after the addition of rYB-1. Here, cell proliferation was blocked by the addition of mitomycin C before cell stimulation to monitor cell migration exclusively. rYB-1 markedly increased cell migration compared with the basal migration rate under unstimulated conditions, to a similar extent as observed with FCS (Fig 3A). This effect was absent when denatured rYB-1 protein was used, excluding the effects by LPS contaminants (Fig 3B). These results are reminiscent of the findings using HMGB1, which induces cell migration in a similar manner (Degryse et al, 2001). To address the question whether full-length YB-1 or the subdomains confer them the extracellular functions, oligopeptides (20 amino acids in length) corresponding to various YB-1 epitopes were tested in similar experiments. The biological activities of such short oligopeptides have been shown before using fragments of amino-acyl transfer RNA synthetases (Wakasugi & Schimmel, 1999). A peptide corresponding to epitopes in the cold-shock domain significantly induced cell migration in human mesangial cells (Fig 3C), whereas a control peptide had no effect. Furthermore, the YB-1 peptide induced cytoskeletal changes of human mesangial cell intracellular stress fibres. These findings are also in line with observations made using HMGB1, which can induce stress fibres in cells (Degryse et al, 2001).

In summary, extracellular YB-1 exerts a potent proliferative response in different cell types and promotes cell migration. Domains in the cold-shock protein probably mediate the effects, as synthetic peptides mimic such functions. This raises the question whether the observed protein fragments (compare supplementary Figs 1 and 2 online) are functionally active.

Given the striking similarities between YB-1 secretion and other non-classically secreted proteins of the nucleus (HMGB1 and heat-shock proteins), one might speculate that secretion of highly conserved nuclear proteins reflects an evolutionarily conserved mechanism to induce inflammation and cell activation.