|Home | About | Journals | Submit | Contact Us | Français|
The pathological isoform of the prion protein (PrPres) can serve as a marker for prion diseases, but more practical tests are needed for preclinical diagnosis and sensitive detection of many prion infections. Previously we showed that the quaking-induced conversion (QuIC) assay can detect sub-femtogram levels of PrPres in scrapie-infected hamster brain tissue and distinguish cerebral spinal fluid (CSF) samples from normal and scrapie-infected hamsters. We now report the adaptation of the QuIC reaction to prion diseases of medical and agricultural interest: human variant Creutzfeldt-Jakob disease (vCJD) and sheep scrapie. PrPres-positive and -negative brain homogenates from humans and sheep were discriminated within 1–2 days with a sensitivity of 10–100 fg PrPres. More importantly, in as little as 22 h we were able to distinguish CSF samples from scrapie-infected and uninfected sheep. These results suggest the presence of prions in CSF from scrapie-infected sheep. This new method enables the relatively rapid and sensitive detection of human CJD and sheep scrapie PrPres and may facilitate the development of practical preclinical diagnostic and high-throughput interference tests.
Transmissible spongiform encephalopathies (TSEs) or prion diseases are neurodegenerative diseases that affect a wide range of mammalian species including sheep (scrapie) and humans [e.g. variant Creutzfeldt–Jakob disease (vCJD)]. A fundamental pathogenic event in prion diseases is the conversion of the normal host prion protein (PrPsen) into a misfolded isoform known as PrPres or PrPSc (Caughey et al, 2009). PrPres appears to be the primary component of the TSE infectious agent, or prion.
A variety of cell-free reactions have been developed to investigate PrP conversion and prion propagation (Kocisko et al., 1994; Bessen et al., 1997; Horiuchi et al., 1999; Saborio et al., 2001; Wong et al., 2001b; Zou and Cashman, 2002; Castilla et al., 2005; Deleault et al., 2005, 2007; reviewed in Caughey et al., 2009). These reactions have shown that PrPres induces the conversion of PrPsen into a PrPres-like protease-resistant state, apparently via a seeded or templated polymerization mechanism (Gadjusek, 1988; Jarrett and Lansbury, 1993; Caughey et al., 1995; Horiuchi et al., 1999, 2000; Castilla et al., 2006). The protein misfolding cyclic amplification (PMCA) reaction can detect as little as ~1.2 ag of PrPres in multi-round sonicated reactions using brain-derived PrPsen as a substrate (Saa et al., 2006). The use of bacterially expressed recombinant PrPsen (rPrPsen) can improve the speed and practicality of PMCA assays (Atarashi et al., 2007). Further simplification can be achieved by substituting shaking for sonication as described for quaking-induced conversion (QuIC) reactions developed for the detection of hamster scrapie PrPres (Atarashi et al., 2008). Within 1 day, the QuIC assay can detect sub-femtogram amounts of PrPres (less than one lethal intracerebral dose) in hamster brain homogenates and can discriminate scrapie-infected and normal hamsters using 2 μl samples of cerebral spinal fluid (CSF).
Here we describe the adaptation of the QuIC assay to PrPres detection in species of greater medical and economic importance than hamsters: humans and sheep. Our findings demonstrate that QuIC assays can amplify femtogram levels of PrPres in both human and sheep brain tissues. Furthermore, our results show that this technique allows the detection of PrPres in peripheral tissues such as CSF from scrapie-positive sheep. These assays may facilitate diagnostic screening of biological samples expected to contain low levels of PrPres such as blood, CSF and lymphoid tissues.
Expression and purification of hamster, human and sheep recombinant PrP molecules (rPrPsen) were done as described previously (Atarashi et al., 2007, 2008) with slight modifications: the protein was eluted using a linear gradient from 0 to 100% elution buffer (500 mM imidazole, 10 mM Tris, 100 mM sodium phosphate, pH 5.8) and collected in 3–4 volumes of pre-chilled 10 mM sodium phosphate dialysis buffer, pH 5.8. The purity of the final protein preparation was ≥99%, as estimated by SDS–PAGE, immunoblotting and mass spectrometry (data not shown).
Sheep or human brain homogenates (BH, 10%) were prepared as described previously (Saa et al, 2006), divided into small aliquots and kept at −80°C. Positivity for natural sheep scrapie and vCJD disease was confirmed by both western blot and immunohistochemistry. Sheep CSF samples were divided into small aliquots and kept at −80°C.
For QuIC assays, we prepared first and second round reactions as described (Atarashi et al, 2008) with minor modifications. When making brain homogenates dilutions, to ensure even distribution of PrPres, a solution of 0.1 or 1% SDS in PBS with the addition of N2 supplement (Invitrogen) was used for human and sheep BH dilutions, respectively. For the proteinase K (PK) digestions of the reaction products, final PK concentrations 3 and 7.5 µg/ml were used for reactions seeded with sheep and human samples, respectively. Immunoblotting of the products was done as described previously (Atarashi et al., 2008). We probed the membrane with 3F4 (Kascsak et al., 1987) or R20 (Caughey et al., 1991) primary antibodies diluted 1:3000 and 1:20 000, respectively, and visualized antibody binding by the Attophos AP Fluorescent Substrate system (Promega) according to the manufacturer's recommendations.
To optimize reaction conditions for vCJD and sheep scrapie-seeded conversions, we varied temperature, shaking, and incubation time, as each of these variables can influence the rate and efficiency of QuIC reactions (Atarashi et al., 2008).
For the detection of vCJD PrPres (PrPvCJD), we initially used human rPrPsen (hu-rPrPsen), (residues 23–231) as a substrate. Reactions seeded with dilutions of vCJD brain homogenates (MM homozygous at codon 129) were first incubated with periodic shaking for 12 h at 45°C, treated with PK and subjected to immunoblotting using an anti-PrP antibody (3F4) directed against residues ~109–112. We observed no PK-resistant recombinant PrP (rPrPres) products in these first-round reactions (not shown). Therefore, we performed a second-round QuIC reaction by diluting an aliquot of the first-round reaction 1:10 into fresh conversion buffer containing hu-rPrPsen. In this round we used a higher temperature (50°C) to promote further amplification of conversion products from the first reaction. Here we detected variable amounts of a 17-kDa rPrPres product in replicate reactions seeded with 3.5 × 10−6 and 3.5 × 10−7 vCJD BH dilutions (Fig. 1A). According to semiquantitative immunoblots using purified hamster PrPres as standards (not shown), these seed dilutions contained ~100 and 10 fg of PrPvCJD, respectively, which are several orders of magnitude lower than the detection limit (~1 ng) of our immunoblots. Thus, the vCJD-seeded 17-kDa rPrPres product (hu-rPrPres(vCJD)) was derived from the hu-rPrPsen substrate rather than the seed. Moreover, this hu-rPrPres(vCJD) resembled the 17-kDa rPrPres product of hamster scrapie-seeded QuIC reactions observed previously using hamster rPrPsen as conversion substrate (Atarashi et al., 2008). In contrast, BH from an Alzheimer's disease (AD) patient failed to induce rPrPres formation, indicating specificity of the QuIC assay for vCJD versus another neurodegenerative disease that involves the accumulation of an amyloidogenic protein. Under these reaction conditions, and a variety of other conditions that we tested using hu-rPrPsen as a substrate (not shown), we found that two reaction rounds were required for detection of PrPvCJD in the femtogram range.
As an alternative strategy, we tested whether vCJD PrPres could induce the conversion of hamster rPrPsen (ha-rPrPsen), (residues 23–231) based on previous successes in using this substrate in single-round QuIC reactions (Atarashi et al, 2008). In these experiments, we analyzed the products using a ha-PrP reactive antiserum (R20) directed at C-terminal residues 218–232, which our previous studies showed can facilitate the discrimination of scrapie-seeded and unseeded QuIC reaction products (Atarashi et al, 2007, 2008). When using this antiserum, hamster scrapie-seeded QuIC and rPrP-PMCA reactions yield rPrPres(Sc) bands of ~11, 12, 13 and 17 kDa (Atarashi et al., 2007, 2008). In contrast, reactions seeded with normal hamster BH yield either no PK-resistant products or a spontaneous PK-resistant product (rPrPres(spon)) that has little, if any, of the 17 kDa band, and a set of 10, 11 and 12 kDa bands that as a set appear down-shifted by ~1 kDa when compared with the rPrPres(Sc) bands. Surprisingly, variant CJD-seeded QuIC reactions with hamster rPrPsen could also give rPrPres(Sc)-like conversion products which we call rPrPres(vCJD). Such products were seen in all 12 h, 45°C QuIC reactions seeded with 3.5 × 10−6 (n = 6) and 3.5 × 10−7 (n = 6) dilutions of vCJD BH (containing ~100 and 10 fg PrPres, respectively) and 2/6 (two out of six) reactions seeded with a 3.5 × 10−8 dilution (containing ~1 fg PrPres) (Fig. 1B top, lanes 19 and 21). In a minor subset of reactions seeded with either vCJD or AD BH samples, a rPrPres pattern consisting predominantly of an ~11 kDa band was observed at this reaction temperature (exemplified by lanes 20, 22 and 23 in Fig. 1B top and lanes 2 and 6 in Fig. 1B bottom). We interpret the latter as a spontaneously generated rPrPres product (rPrPres(spon)) that is not seeded by prions. Rarely (e.g. in 1/62 of the AD BH-seeded reactions performed under similar conditions, 40–50°C, 8–18 h), an apparent combination of the two patterns was observed as shown in lane 4 of Fig. 1B (top) and, in a vCJD-seeded reaction, in lane 7 (bottom); we are uncertain how to interpret this intermediate pattern, which, in the AD BH-seeded reactions, may have been due to inadvertent artifactual prion contamination. At present, we assume that such a pattern implicates the possible presence of prions which should be confirmed by retesting. In any case, the PrPCJD sensitivity for eliciting rPrPres(vCJD) in the 12 h, 45°C assay was in the 1–10 fg range. These reaction conditions provided more consistent and sensitive detection of vCJD than we observed in the 8 h reactions at 50°C (Fig. 1B, lower panels) which had a greater tendency to include overlapping rPrPres(spon) banding patterns (e.g. Fig. 1B, lane 7, bottom). The latter are likely promoted by the higher reaction temperature (Atarashi et al., 2008). Collectively, these results showed that under these two conditions, hamster rPrPsen was a better substrate for the vCJD-seeded QuIC reactions than human rPrPsen, allowing for a more efficient amplification and consistent detection in single-round reactions.
In our previous work we have shown that at any given temperature, shorter QuIC incubation periods can decrease rPrPres(spon) formation (Atarashi et al., 2008). Thus, to reduce the frequency of rPrPres(spon) formation in our QuIC reactions with ha-rPrPsen substrate, we tried reducing the reaction time from 12 to 10 h in the 45°C reactions. vCJD BH's diluted 3.5 × 10−7 (~10 fg PrPres) and 3.5 × 10−8 (~1 fg PrPres) each generated rPrPres(vCJD) in 3/3 reactions (Fig. 1C). In contrast, in three separate experiments conducted at 45°C for 10 h, each with three to six replicates, reactions seeded with AD BH dilutions generated no rPrPres. These results confirmed that reducing the incubation time decreased the incidence of rPrPres(spon) generation under these conditions, but allowed discrimination of vCJD and AD brain homogenates.
To adapt the QuIC reaction to the detection of sheep scrapie PrPres, we initially tested full-length sheep ARQ and VRQ [representing natural amino acid polymorphisms at codons 136, 154 and 171 as determined by full-length sequencing of the ORF, respectively (Hunter, 2007)] rPrPsen substrates in sheep scrapie-induced QuIC reactions. Although we observed the formation of PK-resistant products, we were unable to reliably discriminate between scrapie-seeded samples and negative controls (data not shown). Because hamster rPrPsen was an efficient substrate in vCJD-seeded QuIC reactions as shown above, we tested whether the same might be true in sheep scrapie-seeded QuIC reactions. We tested both ha-rPrP (23–231) and truncated ha-rPrP (residues 90–231) in these experiments and observed no difference in their ability to serve as a substrate for the reaction. We seeded these reactions with both ARQ and VRQ PrPres from two different breeds, Cheviot and Sarda. In the first set of experiments uninfected ARR/ARR BH's were used as a negative control due to our initial lack of uninfected ARQ and VRQ BH's. Samples were incubated at 40°C for 24 h and ha-rPrPsen (90–231) was used as a substrate. We detected a rPrPres(Sc)-like product in all (n = 3) reactions seeded with a 4.3 × 10−5 dilution of 10% VRQ BH containing ~100 fg of PrPres (Fig. 2A). In contrast, in negative control reactions seeded with a 4.3 × 10−4 dilution of ARR BH, no rPrPres was observed in two replicate reactions (Fig. 2A). When using ha-rPrPsen (23–231) as substrate, under the same reaction conditions, we observed rPrPres(Sc) formation in all (n = 3) reactions seeded with 3.8 × 10−6 dilution of ARQ BH (~100 fg of PrPres) (Fig. 2B). In contrast, the reactions (n = 3) seeded with a 3.8 × 10−5 dilution of normal ARQ BH showed no rPrPres formation. This level of sensitivity (~100 fg) was confirmed in five independent experiments in which both ARQ and VRQ scrapie-infected BHs were used as seeds.
To enhance the sensitivity of the sheep scrapie-seeded ha-rPrPsen assay, we performed multiple round reactions. ARQ normal BH was used as a negative control for comparison to ARQ scrapie BH. After a first round at 42°C (14 h), we performed a second round at 45°C (6 h) to promote further amplification of initial conversion products. We detected rPrPres(Sc) in both reactions seeded with a 3.8 × 106 dilution of ARQ BH (~100 fg of PrPres) and in all (n = 3) reactions seeded with 3.8 × 10−7 dilution (~10 fg of PrPres) of ARQ BH, thus improving the sensitivity of the assay by ~10-fold (Fig. 2C). These results were confirmed in two independent experiments.
After assessing the maximal level of sensitivity of our sheep scrapie-ha-rPrPsen assay, we attempted to detect PrPres in a fluid that is more accessible than brain tissue, i.e. CSF. We seeded our initial reactions with 5–10 µl of neat CSF from an ARQ/ARQ scrapie-positive sheep. Negative controls were seeded with equivalent volumes of CSF deriving from a scrapie-negative VRQ/ARQ sheep. Ha-rPrP (23–231) was used as a substrate. Samples were incubated at 45°C (12 h) for the first round and no rPrPres(Sc) product was observed (not shown). However, following a second round at 45°C (Fig. 3A) or 50°C (Fig. 3B), distinct rPrPres(Sc) conversion products displaying variable intensities were observed in all scrapie CSF-seeded reactions. Thus, we proceeded to a third-round reaction. When samples were incubated at 45°C (Fig. 3C), all scrapie CSF-seeded reactions showed an increase in rPrPres(Sc) product formation when compared with the second round. Interestingly, when using a higher temperature (50°C), scrapie CSF-seeded reactions displayed even stronger rPrPres(Sc) products (Fig. 3D), while negative control reactions showed little evidence of rPrPres(spon) formation.
Next, we tested a blinded panel of CSF samples from nine scrapie positive (two ARQ/ARQ, four ARQ/VRQ, two ARH/VRQ and one ARQ/ARH genotype) and three scrapie-negative sheep (two VRQ/ARR and one VRQ/ARQ genotype). The key was broken only after all the data were collected. Reactions were seeded with 10 µl neat CSF, incubated at 45°C (12 h) and ha-rPrP (23–231) was used as a substrate. Consistent with our previous results, no rPrPres was observed after the first round (not shown). Following a second round (Fig. 4A), rPrPres(Sc) was observed in at least one of the duplicate reactions seeded with CSF from seven of the scrapie-positive sheep. These products were further amplified in a third round (Fig. 4B) at 50°C (10 h), allowing detection of rPrPres(Sc) in one additional reaction seeded with a scrapie-positive CSF sample. CSF samples from one of the scrapie-positive sheep produced rPrPres(Sc) in only one of the two replicates (Fig. 4B). Furthermore, in this particular experiment, CSF from one scrapie-affected sheep produced no rPrPres(Sc) in either of the duplicates. However, in a previous experiment, the same CSF sample produced rPrPres(Sc) in two of the four 3-round replicate reactions (data not shown). The need for two to three QuIC rounds and the variability observed in between some replicate reactions suggest that the PrPres seed in scrapie sheep CSF samples is close to detection limit for the assay. This indicated the importance of analyzing multiple replicates of sheep CSF samples. Importantly, in reactions seeded with CSF from three normal sheep, we observed either little detectable rPrPres or a PrPres(spon) banding pattern that was clearly distinct from the rPrPres(Sc) pattern generated in most of the scrapie CSF-seeded reactions. These results, which were confirmed in two independent experiments, demonstrate the ability of our sheep scrapie-ha-rPrPsen QuIC assay to detect PrPres in a biological sample that can be collected from live animals.
A fast and sensitive assay for PrPres detection might have applications not only in basic research, but also in the pre-clinical diagnosis of TSEs and the identification of potential sources of infection. The level of PrPres associated with an infectious unit of prions can be quite low (e.g. ~1–10 fg in the case of i.c.-inoculated hamster 263 K scrapie). Thus TSE diagnostic tests based on the detection of PrPres need to be extremely sensitive to rule out the presence of infectivity. We have now shown that vCJD- and sheep scrapie-seeded QuIC reactions with hamster rPrPsen substrate can detect 10–100 fg of PrPres in as little as 1–2 days. When comparing the sensitivity of the sheep scrapie assay with the vCJD reaction, the latter proved to be more efficient, allowing the detection of ~10 fg of PrPres (3.5 × 10−7 dilution of vCJD BH) in a single-round reaction. In the sheep assay, longer incubation periods were required to detect 10-fold more PrPres (~100 fg or 3.8 × 10−6 dilution of scrapie BH) in a single round. However, a two-round reaction allowed a sensitivity equivalent to the vCJD-ha-rPrPsen QuIC (~10 fg).
Using BH as a substrate, the PMCA reaction has detected PrPres from a 10−5−10−6 dilution of 10% vCJD BH (the PrPres content was not reported) in 24 h reactions and from a 10−7 dilution in 48 h (Jones et al., 2007, 2009). By comparison, our QuIC assay detected 10−7−10−8 dilutions of 10% vCJD BH in 10 h reactions. Adaptation of the PMCA to sheep scrapie allowed amplification of PrPres from a 10−8 dilution of 10% VRQ scrapie BH (quantity of PrPres in BH was not reported) in a 72 h reaction but was less sensitive with the ARQ genotype (Thorne and Terry, 2008). In comparison, our sheep scrapie/ha-PrPsen assay detected PrPres from 10−6 to 10−7 10% sheep BH dilution in 20–24 h reactions, independent of the host genotype and prion strain tested. It should be noted that comparisons with assay sensitivities reported in the literature on the basis of BH dilutions are confounded by variations in the concentration of PrPres among individuals and brain regions sampled. Moreover, our estimates of PrPres content in our sheep and human BH samples were based on immunoblotting against purified hamster PrPres standards and it is possible that the antibodies used for detecting sheep and human PrPres, i.e. the 505 and 3F4 antibodies, respectively, were not equally sensitive in detecting the hamster PrPres standards. Thus, our estimates of the PrPres sensitivity of the QuIC assays must be regarded as only approximate.
The reason that hamster rPrPsen was a more effective substrate than the homologous human or sheep rPrPsen in these reactions is unclear. It is possible that sequence incongruities between seed and substrate become less influential than usual under the QuIC reaction conditions, which include SDS and elevated temperatures. These conditions should partially unfold the substrates and may relieve structure-dependent restrictions on seed–substrate interactions. Under those circumstances, it is possible that the various rPrPsen molecules exist in multiple conformational states and that only one of those states can interact effectively with seeds, as described previously for the amyloid seeding assay (Panza et al., 2008; Stohr et al., 2008). We speculate that the hamster rPrPsen substrate may be better at populating the conversion-competent state and avoiding off-pathway interactions than the sheep or human rPrPsen molecules under the QuIC reaction conditions.
It is important to emphasize that both spontaneous and prion-seeded conversion of rPrPsen molecules to protease-resistant forms can be sensitive to reaction times, numbers of rounds, and conditions, including differences in non-prion components of test samples. Hence, in order to establish prion assays with useful levels of accuracy, it is critical to carefully assess these effects with each type of sample and to include replicates and negative and positive controls in each experiment. Optimally, conditions can be found under which prion-seeded conversion occurs rapidly but spontaneous, prion-independent, off-pathway rPrPsen aggregation and rPrPres formation is rare. Moreover, the higher the sensitivity of an assay, the greater the chance that inadvertent prion contamination of tissue samples, reagents or reaction vessels could cause false positives. Thus, care must be taken to minimize or eliminate such problems and to confirm individual positive reactions with replicates and/or independent tests.
An important characteristic of a TSE test is specificity for prion diseases versus other neurodegenerative diseases. Importantly, both the human intra-species and cross species QuIC reactions amplified PrPvCJD but did not detect the β-amyloid that is present in Alzheimer's brain tissue, except perhaps for a single reaction (out of 62) that gave the equivocal rPrPres pattern shown in Fig. 1B, lane 4 (top). Further studies will be required to determine QuIC assay specificities with respect to other human and sheep disease states.
Another desirable attribute of prion tests is applicability to multiple prion variants. So far we have shown that ARQ- and VRQ-derived PrPres from different geographical origins and sheep breeds can seed our QuIC reactions. However, extensive additional testing will be required to determine the detectability of the many other known types of sheep prions including sheep and goat bovine spongiform encephalopathy (BSE).
One appealing possible application of the QuIC assay is the screening of accessible fluids such as blood and CSF. Different approaches have been taken to detect PrPres in human and sheep CSF. Bieschke et al. (2000) detected PrPres in CSF samples from CJD-positive patients by using a technique called Scanning for Intense Fluorescent Targets (SIFT). However, they could only detect a specific high-intensity signal in 5 out of 24 samples from patients with probable or definite CJD. Wong et al. (2001a) were unable to detect differences in protease-resistance or conformation of PrP in CSF from CJD-positive and non-CJD CSF samples by immunoblotting or conformation-dependent immunoassay. These findings suggest that, as might be expected, the amount of PrPres in human CSF is extremely low and a very sensitive assay is required to detect it. In the case of sheep scrapie, Hadlow et al. (1982) have shown that the level of infectivity in scrapie-positive sheep CSF can range from undetectable to 102.2 LD50 per 30 mg of tissue when inoculated intracerebrally in wild-type mice. Picard-Hagen et al. (2006) reported a slightly different PK sensitivity of PrP in normal versus scrapie CSF, but no PrPres with PK resistance typical of scrapie brain-derived PrPSc was detected. In contrast, our sheep scrapie/ha-rPrPsen assay detected PrPres in CSF samples from all (n = 10) scrapie-positive sheep (although not all replicate reactions were positive) and none of the control sheep (n = 4). Considering that we were able to detect ~10 fg of PrPres in scrapie-infected BH dilutions after a two-round reaction, the fact that a two- to three-round reaction is required to discriminate between scrapie-positive and -negative CSF samples may imply that scrapie-positive CSF contains <10 fg/μl of PrPres. Such low levels might be consistent with the Hadlow et al. bioassay results, but the quantitative relationship between sheep scrapie PrPres and LD50 as measured in mice is unknown. In any case, the ability of our sheep scrapie-ha-rPrPsen assay to identify scrapie-positive CSF samples in as little as 22 h suggests the potential applicability of this technique for the recognition of scrapie infections. In addition, because natural scrapie in sheep has many pathological features in common with human vCJD (i.e. comparable spread of PrPres in peripheral tissues), our results in the sheep model suggests that a QuIC assay might be useful for detecting PrPres in CSF from vCJD-positive humans.
In summary, QuIC is a relatively fast assay that can detect low femtogram levels of PrPres in both human vCJD- and sheep scrapie-infected brain tissue and can identify CSF specimens derived from scrapie-infected sheep. This method could be used on its own or as a front-end amplification step to overcome sensitivity limitations in currently used TSE diagnostic tests. The use of alternative methods to detect the QuIC amplification product (e.g. fluorescently tagged rPrPsen) might further enhance the practicality of this assay.
This research was funded by the Intramural Research Program of the NIAID, NIH. C.D.O. was partially supported by the Master and Back Program of the Regione Sardegna (Italy), J.M.W. was supported in part by the Undergraduate Scholarship Program of the NIH.
The authors wish to thank Dr. Rick Race (Rocky Mountain Laboratories, NIAID, NIH) for the brain samples collected from Cheviot sheep, Drs. Pierluigi Gambetti and Kay Edmonds (Case Western Reserve University, National Prion Disease Pathology Surveillance Center) for providing the human tissue samples. We appreciate the critical reading of this manuscript by Drs. Gerald Baron, Suzette Priola and Kelly Barton. We are indebted to Greg Raymond for his assistance in coordinating the transfer of tissue samples between institutions.
Edited by Neil Cashman