|Home | About | Journals | Submit | Contact Us | Français|
Protein aggregation is associated with the pathology of many diseases, especially neurodegenerative diseases. A variety of structurally polymorphic aggregates or preaggregates including amyloid fibrils is accessible to any aggregating protein. Preaggregates are now believed to be the toxic culprits in pathologies rather than mature aggregates. Although clearly valuable, understanding the mechanism of formation and the structural characteristics of these prefibrillar species is currently lacking. We report here a simple new approach to map the nature of the aggregate core of transient aggregated species directly in the cell. The method is conceptually based on the highly discriminating ability of aggregates to recruit new monomeric species with equivalent molecular structure. Different soluble segments comprising parts of an amyloidogenic protein were transiently pulse-expressed in a tightly controlled, time-dependent manner along with the parent aggregating full-length protein, and their recruitment into the insoluble aggregate was monitored immunochemically. We used this approach to determine the nature of the aggregate core of the metastable aggregate species formed during the course of aggregation of a chimera containing a long polyglutamine repeat tract in a bacterial host. Strikingly, we found that different segments of the full-length protein dominated the aggregate core at different times during the course of aggregation. In its simplicity, the approach is also potentially amenable to screen also for compounds that can reshape the aggregate core and induce the formation of alternative nonamyloidogenic species.
Accumulation of macroscopically observable, abnormal protein deposits is the cytological feature of many age-related neurodegenerative diseases, including Alzheimer’s, Parkin-son’s, and Huntington’s diseases (1–3). Although the characteristic pathological aggregates are primarily made up of the fibrillar aggregated form of the disease-associated protein, multiple aggregate morphologies are accessible to each individual amyloidogenic protein. The heterogeneity of aggregate morphologies may arise from (1) variations in the environment and aggregate growth conditions (4–8), (2) different segments of the primary sequence that govern the formation of the aggregate core (8–13), or (3) complex pathways of formation, including alternate aggregation routes or multistep assembly pathway(s), leading to multiple metastable intermediates (14–20). Current models propose that the species formed early in the course of aggregation, e.g., oligomers and protofibrils, are likely culprits in cytotoxicity and cell dystrophy (21–25). Conformation-specific antibodies that recognize only the oligomers (26) or bind only to the mature fibril state (27), and derived from amyloidogenic proteins of unrelated sequences, point to unique conformational epitopes present only in these states. This suggests that different moieties from a protein’s sequence can be involved in the aggregate core and that the differing nature of intermolecular interfaces might be a rationale for the structural diversity of aggregated species, in turn reflecting the disease phenotype and determining the specific pathological features. Although clearly valuable, details of the structure of transiently formed intermediate species are presently lacking, particularly in the context of a cell, in large measure because of the technical difficulties due to their transient existence.
We have developed an orthogonal cross-seeding assay that can be applied directly to proteins expressed in cells in order to identify sequences or domains involved in the core of aggregates. Amyloid formation generally follows a nucleated-polymerization mechanism. Characteristic of this mechanism is the observation that preformed aggregates or seeds of the same protein can bypass the initiating energetically unfavorable nucleation event (28). Conceptually, the key to our approach is based on the highly discriminating ability of these seeds to recruit new monomers with equivalent sequences (10, 29, 30). The degree of sequence identity determines the efficiency of the seeding process and has been suggested to be crucial for the establishment of the long-range interactions that stabilize the core of the amyloid structures (29). To experimentally probe the nature of the aggregates of an amyloidogenic protein, the model expression host is cotransformed with multiple plasmids, one encoding the full-length amyloidogenic protein and others encoding peptide fragments comprising different parts of the full-length protein (Figure 1A). Provided there is tight control of the promoters, the synthesis of the probe peptides can be turned on and off in a tightly controlled, time-dependent manner. Recruitment of any of the probe fragments into the aggregates of the full-length parent protein will suggest that a portion of the full-length protein sequentially equivalent to the probe fragments comprises the aggregate core.
We used this approach to monitor directly in the cellular environment the structural evolution of the aggregate core of a chimera comprising a well-folded and stable β-sheet protein (cellular retinoic acid binding protein 1, CRABP1) fused N-terminally to exon 1 of huntingtin (Htt) containing a consecutive polyglutamine (polyQ) repeat in the pathological range (53Q), as implicated in the pathology of Huntington’s disease (HD). At least nine slowly progressing hereditary neurodegenerative diseases, including HD, dentatorubral-pallidoluysian atrophy, and several spinocerebellar ataxias, are linked to an expanded unstable polyQ tract in otherwise unrelated disease proteins (3). The disease phenotype is manifest over a threshold value of ca. 35 glutamines, and the polyQ-dependent disease risk and tendency to aggregate are recapitulated both in vitro and in cell culture (31). Here, we show that the core of the aggregates of the CRABP Htt53 chimera evolves in the in-cell aggregation time course: early on, it is formed by the CRABP domain, and later, it is dominated by polyQ segments.
DNA encoding mouse cellular retinoic acid-binding protein I (CRABP) with an N-terminal His-tag and an internal mutation that introduces a tetra-Cys-FlAsH dye binding site (previously described and used to directly monitor aggregation in cells (32)) was cloned under the control of the T7 promoter in the pET16b (AmR, Novagen) vector (33). The C-terminus was extended by the intact sequence of the full length Htt exon 1 with 53 CAG repeats (32). CRABP cloned in the pET16b plasmid was used as a template to introduce the amino acid substitution of Pro39 to Ala, and the resulting DNA encoding P39A CRABP was subcloned in-frame without a stop-codon into the pBAD/myc (AmR, Invitrogen) plasmid. In a subsequent step, P39A CRABP with the C-terminal myc-tag (EQKLISEEDL) was subcloned into pBAD33 (CmR, ATCC 87402) vector under the control of arabinose-inducible promoter. The Htt exon 1 with 20 CAG repeats and with a 10-residue long, N-terminal HA-tag (YPYDVPDYA) was subcloned into the pBAD33 (CmR) behind the arabinose-inducible promoter. DNA encoding β2m protein has been cloned under the T7 promoter in the pET vector (a kind gift from Sheena Radford) (4). E. coli DH5α was used for all genetic manipulations, and all constructs were verified by DNA sequencing.
The full-length CRABP Htt53 (T7, IPTG-inducible promoter, AmR) and either P39A CRABP (arabinose-inducible promoter, CmR) or Htt exon 1 with 20Q (arabinose-inducible promoter, CmR) were cotransformed in E. coli BL21(DE3) host, carrying the DE3 lysogen for high level expression of T7 polymerase. For expression, only freshly transformed cells bearing the double antibiotic resistance were used. A single colony was used to inoculate LB medium containing 100 μg/mL ampicillin or 25 μg/mL chloramphenicol. Protein expression was induced either by adding isopropyl-β-thiogalactopyranoside (IPTG) (to 0.4 mM) for the T7-promoter based constructs or arabinose (to 0.2%) for the BAD promoter-controlled constructs. In the coexpression experiments, expression of the full-length CRABP Htt53 chimera or β2m was induced by adding IPTG for 60, 120, 180, and 240 min after the culture reached OD600 ) = 0.8. IPTG was depleted by centrifuging the cells at 867g for 5 min and washing the cells twice with fresh medium. The cells were resuspended in the same volume of fresh medium supplemented with the corresponding antibiotics, and the expression of the short fragments was induced by adding arabinose (to 0.2%) for 30 min.
Intact aggregates after the pulse-expression of the fragments were isolated from cells as described elsewhere (34, 35). The concentration of the aggregates was determined spectrophotometrically using the 280 value of 21750 M−1cm−1 (34). Aggregates were diluted to 3 μM in 20 mM Tris·HCl at pH 7.5 containing 150 mM NaCl. Two-hundred microliters were applied to a nitrocellulose membrane (Schleicher and Schuell) through a slot-blot manifold and washed three times with 200 μL of the same buffer. Immunodetection was accomplished using the HA (1:3000, Covance, Berkely CA) or myc antibodies (1:5000, Upstate Biotechnology). Visualization was accomplished via peroxidase-conjugated antirabbit or antimouse secondary antibody (Sigma Aldrich). The membranes were treated with 62.5 mM Tris·HCl at pH 6.7 containing 2% SDS and 100 mM β-mercaptoethanol for 30 min at 50 †C in a sealed container to strip the antibodies, rinsed three times with 10 mM Tris at pH 8.0 containing 150 mM NaCl and 0.05% Tween 20, and subsequently reprobed with polyQ antibody MW1 (1:1000 dilution) (Dev. Studies Hybridoma Bank, Iowa, USA), CRABP (1:1000, Abcam), or β2m antibodies (1:5000 dilution) (Strategic Diagnostic, USA).
In previous studies, we observed that the CRABP Htt53 chimera aggregates in a polyQ-length-dependent manner in a bacterial (Escherichia coli) in vivo model and that the aggregates evolved with time, from detergent-labile, small spherical aggregates into detergent-resistant fibrils (32). These morphologically varied aggregates initially could seed ex vivo the elongation of a slow folding variant of CRABP (36) and later the aggregation of Htt exon 1 with 20Q (34). Here, we sought direct in-cell evidence of how the involvement of different portions of the chimeric protein in the core of the aggregate correlated with the observed seeding properties in vitro. The full-length CRABP Htt53 protein was coexpressed in E. coli with either an HA-tagged Htt exon 1 with 20Q or a myc-tagged point mutant of CRABP, consisting each of the domains of the chimera (Figure 1B). The expression of CRABP Htt53 chimera was induced by adding IPTG for various periods and was terminated by depletion of the IPTG-inducer by changing to fresh medium. The T7 promoter is almost completely repressed once IPTG is depleted (Supporting Information Figure 1A,B). Subsequent to turning-off the expression of the full-length chimera, the probe fragments were pulse-induced with arabinose. and their incorporation in the intact CRABP Htt53 aggregates was detected using the immunoreactivity of the corresponding tag. The expression strength of the PBAD promoter ensures a relatively high synthesis pulse and allows a rapid increase of the concentration of either of the probe fragments within the test time of 30 min (Supporting Information Figure 1C). From the earliest time points tested, CRABP Htt53 aggregates efficiently recruited the CRABP species (anti-myc reacting), with the highest activity at the earliest time-point (60 min) (Figure 2). In contrast, the early aggregates showed almost no ability to stimulate polyQ-mediated elongation as evidenced by the lack of recruitment of the Htt exon 1 with 20Q (anti-HA positive species). The later time point CRABP Htt53 aggregates (240 min), for which we have detected a fibrillar phenotype (34), showed the opposite trend. They significantly enhanced the deposition of the polyQ-containing fragments (HA-positive) and were not able to recruit the myc-tagged CRABP mutant (Figure 2). Unlike the pathological variant with 53Q, the Htt exon 1 with 20Q is not aggregation-prone and remains soluble during the entire cycle of expression (data not shown). Therefore, its detection in the aggregates reflects their intrinsic ability to recruit polyQ species. Similarly, while the aggregation-prone mutant of CRABP partitions between the soluble and insoluble fraction at later time points of expression, in the first 60 min after induction, it is only present in the soluble fraction (33). [N.B. Soluble HA-tagged CRABP lacking the point mutation can be recruited into the aggregates with a CRABP core as well. However, the intensity of the resulting signal is lower (data not shown). The point mutation of CRABP used is P39A, which has been previously shown to cause an intermediate with delayed β-barrel closure to be populated and is as a result aggregation-prone (36). The mutant protein is, therefore, more efficiently recruited into the aggregates raising stronger read-out signal.]
This approach is amenable to simultaneous pulse-induction of both probe fragments, provided that they bear different selection markers (as schematically represented in Figure 1A). Simultaneous induction of both probe fragments by adding arabinose (HA-tagged Htt exon 1 with 20Q bearing a kanamycin resistance and myc-tagged P39A CRABP with a chloramphenicol resistance) and subsequent parallel probing of the aggregates with myc or HA-antibodies showed the same trend observed for a single-expressed probe fragment with the CRABP Htt53 chimera (data not shown). In this case, finding only one of the two probe fragments in the CRABP Htt53 aggregates clearly reflects the intrinsic ability of the aggregate to recruit soluble species with related sequence.
This new approach to in vivo analysis of aggregate properties is based on pulse-expression of fragments comprising parts of the full-length aggregating protein, here Htt exon 1 with 20Q and a point mutant of CRABP. In order for this method to be valid, there must be sufficient expression of the probe fragments. The expression levels of both fragments and the full-length CRABP Htt53 chimera were assessed by SDS–PAGE (Supporting Information Figure 2). The use of strong promoters yielded substantial expression in a short time. Longer induction times are not recommended for the probe fragments, both because the morphology of the aggregate species could change within the time span of the expression and because longer expression may lead to aggregation of the probes themselves. The fusion with polyQ in the nonpathological range (CRABP Htt20) does not aggregate in the E. coli expression system (32), and its coexpression under conditions identical to those for the CRABP Htt53 with either of the fragments did not yield any detectable aggregates (data not shown).
The orthogonal cross-seeding experiments developed here strongly suggest that different regions of the full-length CRABP Htt53 chimera are transiently forming the aggregate core at different stages in the aggregation process. Thus, the primary player in the aggregate core evolves from CRABP in the in the early stages into polyQ in the late-forming fibrillar aggregates. Thus, the results presented here are fully consistent with our previous ex vivo limited trypsinolysis analysis, which we used to map the flexibility and accessibility of various regions of the CRABP Htt53 aggregates (34). The CRABP domain was more protected from proteolysis in the early, detergent-labile species, whereas the polyQ segment was less labile in the aggregates isolated at the later time point (34). Interestingly, these results are also entirely consistent with observations from two other laboratories on ataxin-3 aggregation (15, 37), establishing that polypeptide segments involved in the core aggregate structure can change during the time course of aggregation.
We next explored the promiscuity of the orthogonal cross-seeding approach and asked whether aggregates of a sequence-unrelated amyloidogenic protein, β2-microglobulin, which is amyloidogenic and forms fibrils in patients suffering from hemodialysis-related amyloidosis (38), could also recruit the probe fragments. At different time points of expression of the aggregating β2m protein under the control of the T7 promoter (4), we pulse-expressed the probe fragments (Htt exon 1 with 20Q or P39A CRABP) for 30 min. Neither fragment was recruited into the β2m aggregates, arguing that sequence similarity is a prerequisite for recruitment into aggregates (Figure 3).
Control experiments indicated that there were no nonspecific cross-seeding activities of both fragments (i.e., P39A CRABP monomers recruited into aggregates with polyQ-dominated aggregate cores, or vice versa). We prepared seeds from P39A CRABP or Htt exon 1 with 53Q in vitro and used them in cross-seeding reactions in vitro. P39A CRABP aggregates failed to shorten the lag time of aggregating Htt exon 1 with 53Q, and preformed aggregates of Htt exon 1 with 53Q did not seed P39A CRABP aggregation (Supporting Information Figure 3). In addition, the coexpression of both P39A CRABP and Htt exon 1 with 20Q led to no cross-reaction between the two species or formation of aggregates containing both proteins (data not shown). This strongly argues that sequence equivalence to the aggregate core is required for the probe fragment to elongate any of the aggregate species. Promiscuous cross-seeding would suggest substantial heterogeneity of the aggregate core within a single fibril. More broadly, such a result could account for how one amyloidogenic disease might influence or even promote another. The protein deposits, however, are typically dominated by a single primary amyloidogenic protein, which is consistent with the idea that the forces driving amyloid formation are highly discriminating.
Here, we describe an approach for the determination of the active components of the aggregate core of an amyloidogenic protein directly in cells. Our approach offers several advantages over the analysis of aggregates ex vivo (after isolation) or of in vitro grown aggregates: (1) the direct mapping of the time evolution of the nature of the aggregate core in the natural environment increases the likelihood of detecting metastable species whose transient nature hinders their detection ex vivo; (2) the crowded cellular environment may influence the dynamics of the physiologically relevant species, and it is difficult to reconstitute these effects in test tube experiments; (3) the cellular environment is dynamic and actively responds to environmental changes which could introduce new aggregate species or change the aggregation pathway. Finally, this cross-seeding method can be applied to study the active components of aggregates forms of large or unstable proteins whose purification is very inefficient.
We have optimized the orthogonal cross-seeding approach for the E. coli system where readily controlled, rapid, and high expression of the target proteins is straightforward. Nonetheless, this approach is not limited to the bacterial system. It can be transferred to mammalian cells so long as tightly controlled promoters are used, a sufficient amount of the proteins to be tested can be expressed in a short-time, and expression can be turned on and off effectively. In addition to the use of whole domains, we envision that shorter peptides can be also coexpressed to precisely map the residues comprising the aggregate core. The resulting insight into the active regions of an aggregate and the mechanism of recruitment into aggregates are crucial to identifying toxic species and pathological mechanisms. Elucidation of these active species and their mode of action opens new doors for developing therapeutic approaches to ameliorate misfolding and aggregation disorders. Short peptides have been shown to be an attractive therapeutic strategy for the treatment in misfolding diseases (39). The method we have developed has the potential to be used to screen for compounds that modulate the aggregate core and redirect the aggregation process to the formation of alternative, nonamyloidogenic species.
We greatly appreciate the kind gift of the β2m-expressing plasmid by Sheena Radford.
†This project was supported by the SysMO (KOSMOBAC), the DFG (project IG 73/4-1, and the Heisenberg award IG 73/1-1) to Z.I., and by the National Institutes of Health (grant GM027616 and a 2006 NIH Director’s Pioneer Award to L.M.G.).
SUPPORTING INFORMATION AVAILABLE Experimental methods and results about the expression level of the constructs, tightness of the regulation of the T7 and PBAD promoters, and nonspecific cross-seeding abilities of both fragments. This material is available free of charge via the Internet at http://pubs.acs.org.