Generation of double transgenic mice coexpressing UBQLN1 and P497S mutant UBQLN2 proteins
We crossed congenic C57BL/6 Tg mouse lines expressing either human UBQLN1 or the P497S ALS/FTD mutant UBQLN2 cDNA with one another and obtained progeny with all four permutations of the two transgenes (Fig. 1b): Non-Tg, single transgenic UBQLN1, single Tg P497S UBQLN2 and double Tg UBQLN1:P497S UBQLN2 mice. The four genotypes were obtained in the expected Mendelian and sex frequencies. Because expression of both the UBQLN1 and P497S UBQLN2 transgenes are driven by the same Thy1.2 promoter we examined whether expression of the two transgenes differed in single and double Tg mice. Real-time quantitative PCR (RT qPCR) measurement of mRNA expression in 24-week-old animals revealed unaltered expression of the UBQLN1 and UBQLN2 transgenes in double Tg animals compared to single Tg animals (Fig. 1c). Their faithful expression in the double Tg animals enabled us to evaluate how UBQLN1 overexpression affects disease caused by the P497S mutant transgene.
Immunoblotting was used to analyze expression of the two Tg proteins in the four mouse genotypes. To ensure reproducibility of the findings, three different animals of the same age were analyzed for each genotype for this and most subsequent analyses. Expression of the transgenic human UBQLN1 protein was detected using an antibody specific for FLAG, which was used to tag the protein at its N-terminus [26], whereas expression of the P497S mutant human UBQLN2 protein was detected using a UBQLN2-specific antibody [14]. The Tg human UBQLN2 protein is distinguishable from the endogenous mouse UBQLN2 protein because of its smaller size [14]. Immunoblots of brain and SC lysates made from 52-week-old animals revealed expression of the appropriate Tg proteins in the correct genotypes (Fig. 2a, c). Interestingly, the total amount of endogenous and Tg UBQLN2 protein that accumulated in the brain and SC tissues in double Tg animals was lower than in single P497S mice (Fig. 2b, d). We reasoned that the reduction could have stemmed from reduced aggregation and/or accumulation of mutant UBQLN2 protein in the animals.
Differential effects of UBQLN1 overexpression on behavior in double Tg male and female mice
Mice of all four genotypes were tested by measuring their body weight, rotarod performance, and hind limb grip strength from 6 to 52 weeks of age (Fig. 3). Previous studies showed P497S mutant UBQLN2 mice can be discriminated from Non-Tg mice by age-dependent decline in these tests [14]. However, we were mindful that the P497S UBQLN2 line used for the current experiments had a more attenuated ALS-like phenotype compared to the original line. The attenuation occurred sometime during the repeated backcrossing of the original hybrid line with C57BL/6 mice. A similar attenuation in phenotype was also observed after backcrossing the G93A SOD1 mouse model for ALS into the C57BL6 genetic background [22]. Interestingly, the P506T UBQLN2 mouse model that we generated at the same time as the P497S model developed a worse phenotype after similar backcrossing and could not be bred for the current experiments.
All of the tests were conducted bi-weekly for cohorts of at least 5 animals for each genotype and gender (Fig. 3). The tests were stopped at 52 weeks because our preliminary studies indicated it was sufficient time to detect motor neuron loss in the P497S line. The tests revealed different effects of UBQLN1 overexpression that were gender-dependent, regardless of mutant UBQLN2 expression. Because of the complexity of the results, each will be described separately.
The body weight measurements indicated that single UBQLN1 Tg male and female mice were slightly lighter than Non-Tg mice, with a bigger difference seen at older age (~ 6% reduction in both males and females at 52 weeks of age) (Fig. 3a, b). A similar reduction in body weight was seen in Tg mice overexpressing UBQLN1 using the more ubiquitous CMV-actin promoter [17, 24]. By contrast, single P497S UBQLN2 Tg mice were even lighter than the single UBQLN1 Tg mice, with males having ~ 15% and females ~ 10% reduction in weight compared to the Non-Tg animals (Fig. 3a, b). Interestingly, the double Tg mice were the lightest of the four genotypes, with the males having ~ 18% reduction and females ~ 22% reduction in weight compared to Non-Tg animals.
Analysis of the rotarod data revealed a progressive age-dependent decline in latency to fall for all the genotypes, regardless of gender (Fig. 3c, d). However, major differences were found between the two sexes, and therefore their findings are described separately. To simplify the analysis, only the slopes of the rate of the change in performance are compared.
The changes in rotarod performance in the males (as well as for the females) were best fitted by polynomial trend lines. The duration of the latency to fall for single UBQLN1 Tg males was considerably shorter than Non-Tg animals, for all ages tested (Fig. 3c). However, the rate of decline in performance with age was slightly shallower for the UBQLN1 mice. By contrast, the rate of decline in performance for the single P497S Tg males was the steepest of all the genotypes, consistent with the idea that the mice develop MN disease (Fig. 3c). Interestingly, the rate of decline in rotarod performance in double Tg males was the slowest of all the groups, indicating UBQLN1 overexpression improved performance in mice expressing P497S mutant UBQLN2 protein (Fig. 3c).
The rotarod results for the female mice were notably different than for the males. The genotypes with the slowest and fastest rate of decline in performance were the Non-Tg and both P497S-containing genotypes, respectively (Fig. 3d). The declines were most noticeable after 30 weeks of age. The accelerated decline in the mutant P497S compared to the Non-Tg group is consistent with progressive development of motor neuron disease in the former. The smaller decline seen in Non-Tg animals may reflect gradual age-dependent decrease in body function. The single UBQLN1 Tg group had a steeper decline after the Non-Tg animals, but the difference was not very large. The double Tg group showed the most rapid rate of decline over time at early age (before 30 weeks), but after this period the decline became progressively shallow and appeared to slightly improve towards the end of the test period (Fig. 3d).
Analysis of the hind-limb grip strength were similar to the rotarod trends. Again, differences between the genders were apparent and the changes for both genders were again best fitted by polynomial trend lines (Fig. 3e, f). Analysis of the male results revealed single UBQLN1 Tg and Non-Tg mice had the highest grip strength of the four genotypes, both of which increased progressively at a similar rate with age (Fig. 3e). By contrast, the grip strength of single P497S Tg mice was generally lower than the UBQLN1 and Non-Tg animals. Their strength steadily increased till about 30 weeks of age, after which it rapidly declined, eventually decreasing by 25% compared to Non-Tg animals at 52 weeks of age (Fig. 3e). Double Tg mice were quite different. They had the weakest grip strength at early age (15% lower than Non-Tg animals), but this strength steadily increased, eventually surpassing that of single P497S Tg animals (Fig. 3e). At 52 weeks their grip strength was ~ 10% stronger than single P497S mice. These results when combined with the rotarod results strongly suggests that the strength and motor performance of double Tg male mice is superior than single Tg P497S animals.
Analysis of the grip strength for the female cohorts revealed a different pattern than the males. However, the data again showed single UBQLN1 Tg and Non-Tg groups had equivalent and the strongest grip strength of the four genotypes, both of which continuously increased with age (Fig. 3f). The single P497S Tg group had weaker grip strength than the single Tg UBQLN1 and Non-Tg groups for all ages tested. Their strength too increased progressively till about 30 weeks of age when it reached a plateau. Like the males, double Tg females had the weakest grip strength at early age, but their strength increased more rapidly with age than in any other group, till about 36 weeks of age, when it declined more rapidly than any other group (Fig. 3f).
Taken together the behavioral data indicated neuron-specific overexpression of UBQLN1 decreases body weight, irrespective of P497S transgene expression. Furthermore, the tests revealed a more pronounced male-specific benefit of UBQLN1 overexpression in alleviating rotarod and grip strength deficits typically seen in mice expressing the mutant P497S UBQLN2 transgene.
Overexpression of UBQLN1 reduces accumulation of UBQLN2 inclusions and ubiquitinated proteins in the brain of P497S UBQLN2 Tg mice
The four genotypes were examined for signatures of UBQLN2 pathology. Accordingly, brain sections of 52-week-old animals were stained for UBQLN2 inclusions [14]. As expected, single P497S Tg animals contained numerous UBQLN2-positive inclusions spread throughout the brain, with their characteristic high preponderance around the dentate gyrus of the hippocampus (Fig. 4a and Additional file 1: Fig. S1A) [14, 35]. These inclusions were considerably diminished in the double Tg animals. Quantification of the number of UBQLN2 inclusions separated by size (0.1–1 μm and 1–10 μm) in the dentate gyrus, CA1 and cortex regions of the brain revealed far fewer small and large-size inclusions in the three genotypes compared to the single P497S Tg animals (Fig. 4b and Additional file 1: Fig. S1). Intriguingly, the number of the large size inclusions were reduced in double Tg animals to almost the same levels as Non-Tg and UBQLN1 animals. However, some of the smaller size inclusions still persisted in the double Tg mice (Additional file 1: Fig S1B).
To confirm the reduction of UBQLN2 inclusions in mice overexpressing UBQLN1, we stained the brain sections for ubiquitin and p62/SQSTM1 (Additional file 1: Fig. S2A and B), as both proteins concentrate and colocalize with UBQLN2 inclusions in P497S animals [14, 35]. The results revealed, as expected, single P497S Tg animals contained numerous p62- and ubiquitin-positive foci, most of which colocalized with UBQLN2 staining (Additional file 1: Fig. S2A and B). By contrast, these foci were greatly reduced in the remaining genotypes, with the exception of the double Tg mice where smaller p62 inclusions were sometimes seen. Further analysis of whole brain lysates from the animals by a filter-retardation dot blot assay [27, 32] revealed double Tg mice also had reduced retention of UBQLN2 immunoreactivity on the filters compared to single P497S Tg animals (Fig. 4c). The immunoreactivity in the double Tg animals was comparable to that in Non-Tg and UBQLN1 mice, consistent with the idea they have fewer UBQLN2 aggregates (Fig. 4d).
We also probed whole brain lysates of each genotype for changes in UBQLN2, ubiquitin, and p62. Similar to the immunofluorescence findings, the blots for p62 and ubiquitin revealed dramatic reduction in accumulation and reactivity of both proteins in the double Tg mice compared to single P497S Tg animals (Fig. 2a, b). Taken together, these results indicate UBQLN1 overexpression reduces UBQLN2 inclusions in mice expressing the P497S UBQLN2 mutant transgene.
Overexpression of UBQLN1 reduces neuronal loss in the brain of P497S UBQLN2 Tg mice
A major neuropathologic consequence of overexpression of the mutant P497S transgene is loss of neurons in the hippocampus [14]. We therefore assessed whether UBQLN1 overexpression affects neuronal loss. Accordingly, we counted the number of NeuN-positive neurons in the CA1 and dentate gyrus regions in the four genotypes using stereological principles. The counts revealed, as expected, significant neuronal loss in the dentate gyrus in single P497S Tg animals at 52 weeks of age compared to Non-Tg and single UBQLN1 Tg animals (Fig. 5a, b). The quantifications also revealed a trend toward fewer neurons in the CA1 region of the P497S animals, but the reduction was just outside of significance (P = 0.07). We noted a slight trend in reduction of neurons in single UBQLN1 Tg mice in both the dentate gyrus and CA1 regions compared to Non-Tg mice, but the difference was not significant. More significantly, the quantifications revealed double Tg mice had reduced neuronal loss compared to single P497S Tg mice (Fig. 5b). Interestingly, the neuronal counts in the double Tg animals more closely resembled those found in single UBQLN1 Tg animals than in Non-Tg animals. Nevertheless, the results indicate UBQLN1 overexpression reduces neuronal loss in mice expressing the mutant P497S UBQLN2 protein.
Overexpression of UBQLN1 reduces UBQLN2 inclusions and build-up of ubiquitinated proteins in the spinal cord of P497S Tg mice
We next examined if the abrogation of UBQLN2 pathology found in the brain of double Tg mice extends to the spinal cord (SC), since expression of the mutant P497S transgene is associated with motor neuron disease [14]. Repetition of the blots using SC lysates revealed almost identical findings to those seen in the brain: a reduction in the double Tg animals of endogenous and transgenic UBQLN2 accumulation, decreased protein ubiquitination and reduction in p62 levels compared to single P497S animals (Fig. 2c, d). Immunostaining of the SC sections for UBQLN2 also revealed dramatic reduction in the number of UBQLN2 inclusions in the double Tg mice (Fig. 6a). Quantification of the reduction revealed double Tg mice still had significantly fewer small and large size 0.1–1 and 1–10 μm size inclusions compared to the P497S animals, despite having higher number of large size inclusions compared to Non-Tg animals (Fig. 6b).
Because UBQLN proteins function in ERAD, and ALS-FTD mutations interfere with the process [8, 37], we also probed the SC lysates for evidence of ER stress. Accordingly, we probed the samples for alteration in eIF2α phosphorylation (Ser51), an increase of which is reflective of increased ER stress [9, 28] (Fig. 6c). The blots revealed a strong increase in phospho-eIF2α(Ser51) levels in single P497S Tg animals compared to Non-Tg animals, which was considerably attenuated in the double Tg mice (Fig. 6c, d). Single UBQLN1 Tg mice had comparable phospho-eIF2α(Ser51) levels compared to the Non-Tg animals.
Overexpression of UBQLN1 reduces MN loss and TDP-43 pathology in the SC
Expression of the P497S mutant transgene is associated with age-dependent MN loss in the SC [14]. Accordingly, we quantified the number of MN in both male and female mice at 52 weeks of age for the four genotypes (Fig. 7a, b) [6, 14]. Changes in MN number was confirmed by ChAT staining of the sections (Additional file 1: Fig. S3). The quantification revealed little difference between UBQLN1 and Non-Tg animals. However, both male and female single P497S animals had approximately a 50% reduction in MNs (Fig. 7b). By contrast, MN number in double Tg male and female mice were very similar to their Non-Tg counterparts.
Further evidence supporting neuroprotection by UBQLN1 overexpression was found by examining TDP-43 pathology in spinal MN of the mice [14]. These examinations showed TDP-43 staining was restored to the nucleus in the double Tg mice compared to single P497S Tg mice, where many more MNs had displacement and accumulation of the staining in the cytoplasm (Fig. 7c). Single UBQLN1 Tg and Non-Tg animals showed no signs of TDP-43 pathology, as expected.
Restoration of muscle weight and fiber size in double transgenic male UBQLN1/P497S Tg mice
The weight of the gastrocnemius muscle and size of the muscle fibers was determined for the different animal cohorts at 52 weeks of age (Fig. 8). Measurement of muscle weight in male mice revealed an approximately 45% reduction in single P497S animals compared to Non-Tg animals, consistent with the expected weight loss from motor neuron disease in P497S animals (Fig. 8a). Muscle weight in single UBQLN1 mice was similar to Non-Tg animals. Importantly, its weight in double Tg male mice was similar to UBQLN1 and Non-Tg animals. The comparison of muscle fiber diameter in the male mice revealed a similar trend, with reduction of fiber size in single P497S animals and restoration to close to the normal size in double Tg animals (Fig. 8b–d).
Similar comparison of muscle weight and fiber size in the female cohorts revealed an increase in fiber size, but not muscle weight, in the double Tg mice (Fig. 8b–d). Like the males, muscle weight and fiber size were both reduced in the single P497S Tg female mice compared to UBQLN1 and Non-Tg animals. However, muscle weight was little different between double Tg and single P497S Tg female mice. Interestingly, however, muscle fiber diameter was increased in the double Tg female mice compared with P497S animals.
UBQLN1 overexpression partially restores axonal caliber in male cohorts of double Tg mice
We next quantified the number and diameter of axons in the L4 nerve as an independent assessment of MN changes (Fig. 8a). Quantification of the total number of myelinated axons in transverse sections taken through the L4 nerve of 52-week-old animals revealed a near significant reduction in axon number in male P497S single Tg animals (P = 0.06) compared to Non-Tg animals (Additional file 1: Fig. S3A). Conversely, the numbers in double Tg and single UBQLN1 Tg male animals showed little difference compared to the Non-Tg animals (Additional file 1: Fig. S4A). Furthermore, the same comparison in female cohorts, revealed no difference across all four genotypes (Additional file 1: Fig. S4B).
The profile of the diameter of myelinated axons in both male and female mice showed the standard biphasic segregation of small (0–30 μm) and large (40–150 μm) caliber axons (Fig. 8b, c). However, we identified differences between the profiles of male and female mice. First, the peak of large caliber axons in male cohorts differed considerably across the four genotypes. In Non-Tg animals, it was centered at ~ 90–100 μm. By comparison, single UBQLN1 Tg mice centered at 70 μm, while double Tg mice centered at 40–50 μm. Single P497S Tg mice had the most flattened profile, suggesting the loss of large caliber axons (Fig. 8b). Comparison of the axon profile for the female groups revealed more similarity between the single UBQLN1 Tg and Non-Tg genotypes, although subtle differences were apparent (Fig. 8c). By contrast, both single and double Tg P497S animals had very similar profiles, with reduced number of large caliber axons and a pronounced increase in the cluster of small caliber axons (0–25 μm). The lack of alteration in the large caliber axons in double Tg animals suggests UBQLN1 overexpression did not rescue loss of large caliber axons in the double Tg female cohort. Taken together, examinations of the L4 axons provides additional support that overexpression of UBQLN1 can partially rescue MN loss in male mice expressing the P497S mutant transgene. However, the results also suggest that the benefit does not extend to females.