Conversion of Optineurin into an Hsp72-Inducible protein by C-terminal addition of green fluorescent protein

• Introduction
• Methods
• Results
• Discussion

Abstract

Purpose: Optineurin is known to be associated with glaucoma. This protein has often been investigated as a form of fusion with green fluorescent protein (GFP), but there have been few reports describing any undesired effect of the tag on the normal expression of naïve optineurin. We investigated whether GFP tagging potentially does not influence the characteristics of optineurin expression.

Methods: Wild-type and mutant (E50K and M98K) optineurins were fused with GFP or yellow fluorescent protein (YFP) either at their N-terminus or C-terminus. The fusion constructs were loaded onto an adenoviral vector and analyzed by western blot analysis and immunocytochemistry.

Results: In human trabecular meshwork cells, optineurins fused to GFP at their C-terminus (OPTN-GFP) were prone to aggregation and did not fluoresce as brightly as their counterparts fused to YFP did regardless of whether they were wild-type or mutant forms. In addition, their expression was accompanied by the apparent induction of heat shock protein 72 (Hsp72). Furthermore, OPTN-GFP appears to interact with Hsp72 and be co-aggregated into vesicle-like structures scattered throughout the cytoplasm. However, optineurin fused to GFP at its N-terminus was not aggregate and did not induce Hsp72 expression.

Conclusions: It is well-known that misfolded or unfolded proteins are prone to aggregation and, if they are fused with GFP, their chromophores are not fully fluorescent. Additionally, it is also well-known that heat shock response is a key cellular processes against the accumulation of misfolded or unfolded proteins. Therefore, our data suggest that the addition of GFP to the C-terminus of optineurin might convert it into an unfolded or partially folded protein.

Optineurin, which stands for “optic neuropathy inducing,” is a cytosolic protein with 577 amino acids that was formerly named FIP-2 (14.7K-interacting protein 2) and also called NRP (NEMO-related protein) because it shows a homology to the NF-κB essential molecule (NEMO) [1–3]. The protein has multiple binding partners, such as Rab8, huntingtin, myosin VI, transferrin receptors, and autophagic markers. With its binding proteins, optineurin performs various cellular functions, including the maintenance of Golgi organization, vesicle trafficking, regulation of NK-ĸB signaling, and the induction of autophage. Interestingly, mutated forms of optineurin have been found mainly in neurodegenerative diseases, such as normal tension glaucoma and amyotrophic lateral sclerosis, although they are ubiquitously expressed [4]. In previous studies, optineurin has often been expressed as a fusion with green fluorescent protein (GFP), a 27 kDa protein that spontaneously emits green fluorescence when exposed to light in the blue to ultraviolet range [5–17]. To date, several proteins have been investigated as fusions with GFP or other fluorescent proteins in many species, such as yeasts, fungi, fish, and mammals, including human cells, to facilitate their detection, visualization, and localization. The advantages of the fusion proteins are so well-recognized in a variety of investigations that it doesn’t matter if optineurin is investigated as a GFP fusion in a certain organism [18]. Nonetheless, it is necessary to ascertain that the GFP tag has a minimal influence on the characteristics of a naïve protein, such as its own tertiary structure, biologic activity, and subcellular localization.

In the present study, we accidentally found that heat shock protein 72 (Hsp72) was considerably induced in cells expressing optineurin that were fused to GFP at its C-terminus. Here, we report our unexpected finding and discuss its biological significance with respect to the previous works performed on the GFP-fused optineurin protein.

Initially, we were interested in whether cell death can be accelerated by aberrant optineurins having mutations, such as E50K and M98K, that are highly prevalent in patients with glaucoma [1]. As a first step toward investigating the cellular dysfunctions caused by optineurin mutations, we transduced HTM cells with adenoviruses that express wild-type and mutant optineurins as a form of GFP or YFP-fused recombinant proteins. Unexpectedly, however, it was found that wild-type optineurin fused to GFP at its C-terminus (OPTN-GFP) did not fluoresce as brightly as its counterpart fused to YFP (OPTN-YFP) did, even when observed under the FITC filter that is adequate for green fluorescence. Similarly, mutant optineurins fused to GFP at their C-terminus (E50K-GFP and M98K-GFP) were not fluorescent brightly compared to their counterparts (E50K-YFP and M98K-YFP). There was no difference between wild-type and mutant optineurins in terms of fluorescent intensity (Figure 1A). Western blot analysis showed that comparable amounts of optineurin were expressed in all cells except untransduced cells (Figure 1B), demonstrating that a weak fluorescence of OPTN-GFP was not the cause of its low expression.

Quite unexpected was an additional finding that Hsp72 expression was extraordinarily increased in all cells expressing optineurins fused to GFP at their C-terminus (Figure 1B). The finding is not an artifact merely ascribed to viral infections or protein over-expression because the induction of Hsp72 expression was not observed at all in cells expressing optineurins fused to YFP at their C-terminus as well as in cells expressing native optineurin (OPTN) and optineurin with a flag epitope (OPTN-FLAG). Importantly, Hsp72 expression was not significantly increased in cells expressing optineurin fused to GFP at its N-terminus (GFP-OPTN). Hsp72 was not also induced in cells expressing GFP or MYOC-GFP that served as reference proteins (Figure 2). Immunocytochemistry consistently showed that an increase in Hsp72 expression was apparent only in cells expressing optineurins fused to GFP at their C-terminus. It was noticeable that its fluorescence were not as bright as those of other fluorescent proteins without Hsp72 expression (Figure 3).

Immunocytochemistry also showed that OPTN-GFP was easily aggregated, giving rise to the formation of dispersed perinuclear aggregates (Figure 3). A merged image of fluorescence against the contour of HTM cells highlighted its aggregation-prone property. In contrast to OPTN-YFP and GFP-OPTN that showed typical staining for cytosolic proteins, OPTN-GFP was expressed as small speckles scattered throughout the cytoplasm at a moderate level of transduction or aggregated into large vesicular structures at a higher level of transduction (Figure 4). In addition to its bias to aggregation, it appeared that OPTN-GFP was able to interact with Hsp72 and thereby co-aggregate into discrete vesicles with it, resulting in the co-localization of both proteins (Figure 5A). The interaction of OPTN-GFP with Hsp72 was confirmed by immunoprecipitation in which Hsp72 was co-precipitated when OPTN-GFP was immunoprecipitated with an anti-GFP antibody and vice versa (Figure 5B).

We then expressed OPTN-GFP, GFP-OPTN, OPTN-YFP, and MYOC-GFP in ARPE-19 cells to determine whether Hsp72 can be induced by OPTN-GFP in other cells besides HTM cells. The western blot clearly showed that only OPTN-GFP had an ability to induce Hsp72 expression, although its fluorescence was not bright as compared to other constructs (Figure 6A-C). Additionally, similar results were obtained with other cell lines, including RGC-5 cells, HT1080 cells, and primary human skin fibroblast cells (Figure 6D,E). Therefore, it seems that Hsp72 induction by OPTN-GFP is not a phenomenon confined to HTM cells, but an unidentified cellular event caused by the C-terminal addition of GFP.

In many previous investigations, wild-type and mutant optineurins or fragment optineurins were tagged with GFP at their N-terminus (GFP-OPTN) [5,7,8,13,14] or at its C-terminus (OPTN-GFP) [6,9-12,15,16]. The GFP gene is currently originated from a series of pEGFP vectors (Clontech) that encodes enhanced green fluorescent protein (EGFP), a red-shifted variant of wild-type GFP, which contains the double-amino-acid substitution and more than 190 silent base changes for human codon-usage preferences [20,21]. This modification allows the GFP to be optimized for brighter fluorescence and higher expression in mammalian cells.

When a protein is studied that is a recombinant hybrid containing a relatively large polypeptide, a prerequisite must be fulfilled that the fusion partner potentially do not affect the original characteristics of the target protein. In this study, however, we found that the C-terminal addition of GFP to optineurin resulted in a hybrid protein that did not normally fluoresce but strongly induced Hsp72 expression regardless of whether it was in wild-type or mutant form. It has been reported that the fluorescence of GFP fusion is related to the productive folding of the upstream protein domains expressed alone, and heat shock response can be elicited via the accumulation of thermally misfolded proteins [22,23]. Therefore, it could be inferred from our findings that the addition of GFP to the C-terminus of optineurin might convert it into a protein that is not correctly folded.

More important with regards to regarding the abnormality of OPTN-GFP expression is its aggregation. In this study, OPTN-GFP was expressed as aggregates of various sizes from dispersed punctate forms to large vesicle-like structures. Additionally, OPTN-GFP not only induced Hsp72 expression, but appeared to interact with it and co-aggregated into discrete vesicular structures. Since the primary functions of Hsp72 is to act as a molecular chaperone that recognizes and binds to partially folded or misfolded proteins to prevent their intracellular accumulation as aggregated form, the formation of aggregates of a protein in combination with Hsp72 induction by that protein can be a strong indicator of it misfolding or unfolding [24]. Therefore, it is very likely that data obtained with OPTN-GFP might be at least in part an artifact. However, any comment on this matter has not been mentioned in those studies that have been performed with OPTN-GFP. Only Sahlender et al. reported that over-expressed optineurin or fragments of optineurin tagged with GFP, Flag, or myc at their N-terminus or C-terminus did not have the same intracellular localization as endogenous optineurin, but showed an overall vesicular distribution and formed aggregates in the cytoplasm [25]. They suggested that the addition of the GFP, Flag, or myc tags to both terminus of optineurin or even to its fragments alters the protein conformation in some way and masks domains important for intracellular targeting to the Golgi complex. However, they also did not notice Hsp72 induction concomitant with OPTN-GFP expression.

To our knowledge, this is the first report describing induction of Hsp72 expression by a protein tagged with GFP, a phenomenon that is not merely an artifact due to viral infection or protein over-expression, but a general cellular response to the tagged protein. The reason why OPTN-GFP shows a tendency to aggregate and the mechanism by which it induces Hsp72 expression is not intensely investigated in this study. Unraveling it seems like a very difficult task in individual experimental settings, and it exceeds the scope of the present study. However, it will be worth investigating in further studies. In summary, our results suggest that C-terminal addition of GFP to optineurin can convert it into a misfolded or partially unfolded protein. Therefore, it is highly recommended that any conclusion made in the earlier studies using optineurin fused to the N-terminus GFP should be reconsidered.