HspA1A is a stress-inducible seventy-kilodalton heat shock protein (Hsp70) [1
]. This protein is a molecular chaperone that plays critical roles in protein homeostasis and survival of stressed and cancer cells. In order to function as a molecular chaperone, HspA1A interacts with short hydrophobic regions of misfolded proteins in an ATP dependent manner. As a typical Hsp70, HspA1A contains a nucleotide binding domain (NBD) connected to a substrate binding domain (SBD; β-strand subdomain) via a hydrophobic linker, followed by the C-terminus α-helical region [1
]. The NBD binds and hydrolyzes ATP, whereas the SBD binds exposed hydrophobic regions of proteins [3
]. The chaperone performs its function by alternating between its open and closed conformations. This change is regulated by ATP binding and hydrolysis at the NBD and the resulting allosteric conformation change of the protein [3
In addition to its intracellular anti-apoptotic functions, HspA1A also localizes in the plasma membrane (PM) where it is presented on the surface of many different human tumor types resulting in radiation insensitive and metastatically aggressive tumors [5
]. Thus, inhibiting HspA1A’s PM translocation would augment current therapeutics and detail new ones.
Inhibition of HspA1A’s PM localization requires prior knowledge of the molecular mechanism that drives this translocation. However, the mechanism that HspA1A uses to translocate to the PM is largely unknown, because the protein does not contain any known lipid-binding domains or other signals necessary for protein translocation to or embedding in the PM. Nevertheless, multiple reports demonstrated that HspA1A binds to particular anionic lipids and further revealed that HspA1A’s PM localization, anchorage, and surface presentation depends on its interaction with phosphatidylserine (PS) [6
Therefore, inhibiting the binding of HspA1A to this lipid might stop the protein from localizing at the PM and subsequently HspA1A’s cell-surface presentation. However, the regions of HspA1A that mediate the interaction with PS, as well as the relationship between the lipid-binding and chaperone functions, are largely unknown. This knowledge will provide critical information about the unknown regions of HspA1A that bind to PS, thus mediating the protein’s PM localization, and allow us for future interventions to specifically target the lipid binding function of HspA1A as opposed to targeting the chaperone function, which has detrimental side and off-target effects.
To further characterize the interaction of HspA1A with PS and elucidate whether lipid-binding regulates its chaperone function, we determined the effect of lipids on the secondary structure and chaperone function of HspA1A using the wild-type and a mutated form of the protein that is unable to function as a molecular chaperone.
Recent research revealed that HspA1A translocates to the PM and is presented in the cell surface of multiple cancer types resulting in radiation insensitive and aggressive tumors. Many reports revealed that HspA1A is found in association with sphingomyelin/cholesterol rich lipid domains (lipid rafts) [5
]. Therefore, several investigators aim to target the cell surface/PM-localized HspA1A as a novel anti-cancer treatment [5
]. It is also fairly well established that HspA1A’s PM localization depends on its interaction with intracellular PS and other lipids, including globotriaoslyceramide, which is primarily found in lipid rafts [6
]. However, how HspA1A binds to PS, GB3, or other PM lipids remains a mystery because this protein does not contain any known lipid-binding domains or other PM translocation signals. Thus, if we manage to inhibit HspA1A’s interaction with PS (or other lipids), we might be able to augment current therapies by stopping the protein from translocating to the PM while still maintaining its chaperone functions.
In this report, our goal was to advance our understanding of the HspA1A regions that bind to PS by determining whether the regions responsible for the chaperone functions of HspA1A overlap with the lipid-binding regions.
Our experiments verified that the proteins we used bind to lipids as expected based on the published literature [8
] and provided an important new finding that HspA1A binds significantly higher to saturated PS (DPPS) as compared to mono-saturated PS (POPS). This finding suggests that in addition to the ability of HspA1A to preferentially embed to DPPS containing liposomes, the protein also has a higher affinity for saturated PS species. These observations might have important physiological implications because several well-controlled lipidomics studies revealed an increase in saturated lipid species in stressed and cancer cells [26
]. In addition to the WT HspA1A proteins we also determined that K71A, an ATPase-defective HspA1A mutation, binds to POPS liposomes. We further evaluated this initial finding using SPR for both forms of proteins. The finding that K71A binds specifically to lipids indirectly supports the prediction [8
] that lipids and nucleotides are not accommodated by the same protein sites, because if they did the expectation would be the loss of a positive amino acid in the region to result in loss of lipid-binding. Furthermore, the comparison between the binding characteristics of the WT and the K71A proteins, provided new information and basic clues on the interaction between HspA1A and POPS liposomes as well as the basis for explaining the effect of the K71A mutation on the lipid-binding properties and putative sites of the chaperone. Based on the results for both protein forms, the model that best describes the interaction of HspA1A with POPS liposomes assumes a two-step interaction. During the first step, HspA1A interacts with the liposomes. During the second step, the initial complex formed undergoes a conformational change, which accelerates the binding reaction. This Two state model is in agreement with the binding model proposed to explain the binding of HspA1A to DPPS, BMP, and phosphoinositides [10
] using the LVS method. Furthermore, the proposed mechanism is in accordance with the well-established ability of HspA1A to change conformations after binding to nucleotides or other proteins (co-chaperones or client proteins) and allosterically communicate information between the two major domains of the protein. It is important to note here that under the conditions used HspA1A interacts peripherally with the POPS liposomes [8
], thus the mechanism suggested above does not include the final step of this interaction, which is limited to saturated lipids, the protein embedding in the lipid bilayer [8
]. Although this model is an attractive interpretation of the current literature, it can only serve as an additional indication of conformational changes of HspA1A upon binding to POPS, and warrants further investigation.
In addition to the binding model, the comparison of the kinetics between the WT and K71A protein forms provide further information on the lipid-binding properties of HspA1A. Although the maximal binding capacity and the dissociation equilibrium constant are significantly different between the WT and K71A in agreement with the LVS assay, these values are less informative and the effect size of the change much less pronounced than the kinetics of the interaction. Analysis of the kinetics reveals that the K71A mutation results in a significantly faster association of HspA1A with POPS and very fast protein-lipid dissociation. In contrast, the K71A mutation results in significantly slower generation of the final HspA1A-POPS complex (the product of the presumed conformational change) and slower conversion back to the original complex. Taking into account that the lysine residue in position 71 is buried within the ATP binding cleft, a plausible explanation of these observations could be that the mutation slows the protein’s conformational changes due to the binding to lipids. This idea is indirectly supported by the fact that the K71A mutation results in a protein that does not undergo a conformational change or release peptide in the presence of ATP, although the protein still binds to the nucleotide [16
Several reports have revealed that HspA1A embedding in the lipid bilayer results in oligomerization of the protein [11
], which by changing the quaternary structure of the complex could further support the conformational change concept. Indeed, de-oligomerization of HspA1A results in impairing of clathrin-independent endocytosis and less protein at the cell surface [21
]. To assess whether major changes in the secondary structure of HspA1A occur because of either binding to or embedding into the lipid bilayer we used CD spectrometry. Our results strongly suggest that under the conditions tested lipid binding and embedding do not cause major and specific alterations of the WT protein structure. This finding partially contrasts the consistent loss of the α-helical region when K71A protein was incubated with POPS liposomes. These findings agree with results obtained using bis-(monoacylglycerol)-phosphate (BMP; [28
]) and strongly suggest that although lipids might alter the conformation of HspA1A [8
] they do not seem to affect the overall protein structure. Furthermore, these conformational changes might be very small compared to the binding of nucleotides in the presence of Mg2+
] or transient and very fast. The latter notion is supported by the kinetics of the interaction generated with SPR, which suggest a very rapid turnover for the WT protein and considerably lower transition for the K71A mutation.
Our next set of experiments determined that the addition of liposomes results in a significant increase in ATP hydrolysis. However, all lipids used similarly induced these changes, irrespective of whether HspA1A binds them with high affinity or not. These observations suggest that binding to lipids or embedding in the lipid bilayer does not specifically affect the ATPase activity of HspA1A. These findings are in agreement with previous research showing that the addition of ATP lowers lipid binding to background levels, but this loss-of-binding is less pronounced when HspA1A and lipids are pre-incubated prior to the addition of ATP [8
]. Similar results were observed for another molecular chaperone (Hsp90), which also binds to lipids, although this study did not assess whether this increase is the result of specific lipid-binding [32
]. Another lipid, sulfatide (SGC), induces oligomerization of HspA1A by locking it a high-affinity state to unfolded proteins and blocking the binding to ATP [30
]. However, it is not clear whether this was the result of specific binding or if this was due to the presence of the lipid. Together these observations allow us to speculate that the presence of lipids that HspA1A interacts with, even with low affinities (e.g., PC, the most abundant phospholipid in mammalian cells), might stabilize the protein thus resulting in higher ATP hydrolysis.
The results of the ATPase assay imply that although in structural proximity [13
], the regions that mediate ATP binding and hydrolysis and lipid binding do not overlap. The finding that the K71A mutation, which renders the chaperone unable to hydrolyze ATP, binds to POPS with affinity comparable to the WT’s protein, further supports the idea that ATP hydrolysis and lipid-binding are distinct functions mediated by non-overlapping regions.
Our last set of experiments determined that the addition of liposomes has small effects on the refolding activity of HspA1A. These changes seem to be specific and clearly relate the refolding rate of the chaperone to the binding to specific lipids (i.e., POPS and DPPS). Nevertheless, the effect size of these changes (less than one-fold maximal increase) is relatively small and might not be physiologically important. This notion further strengthens our prediction that the regions of the protein responsible for lipid binding and refolding although related do not overlap. This notion is further supported by the increase in refolding rate, because if the lipid and the protein client competed for the same sites, the expected result would be a decrease in the refolding rate. Another possibility could be that the presence of lipids may directly influence refolding of β-galactosidase as has been described for different membrane proteins [33
]. Although this possibility cannot be formally excluded it seems less probable because most of the refolding increase observed was PS specific and the BSA controls did not show a similar increase. Based on our findings and the current literature, the changes in both the ATPase and refolding activities could be explained by the effect of the lipid in either stabilizing the chaperone, or causing conformational changes that favor ATP hydrolysis and protein refolding. Although attractive, these speculations warrant further experimentation.
In conclusion, our results reveal that similarly to Hsp90, in which the membrane-deforming and chaperone functions are separated [36
], the regions mediating the chaperone and lipid binding functions of HspA1A do not overlap. This knowledge provides new information towards elucidating the lipid-driven mechanism of HspA1A’s translocation to the surface of tumor cells. These mechanistic details will allow us to specifically target the lipid-binding function of HspA1A in order to inhibit the lipid-driven translocation of HspA1A to the cell surface of tumor cells, making them sensitive to radiation therapy.