Galactose-1-Phosphate Uridylyltransferase mRNA Therapy May Improve Movement Problems in a Mouse Model of Classic Galactosemia

mRNA was synthesized and formulated in sterile lipid nanoparticles (LNPs) as described previously [18]. The same biodegradable, ionizable LNPs described in our previous studies was used in the current set of studies [18]. Briefly, sequence-optimized mRNA encoding GALT was synthesized in vitro using optimized T7 RNA polymerase-mediated transcription reaction with complete replacement of uridine by N1-methyl-pseudoridine, following Moderna's proprietary therapeutic protocol. The linearized DNA template contains the 5′ and 3′ untranslated regions (UTRs), open reading frame (ORF), and the poly-A tail. Inverted deoxythymidine (idT) residues were appended to the 3′ terminus of fully synthesized mRNA by phosphodiester linkage of a modifying oligo with 3′ inverted deoxythymidine (3′idT) [18,20]. The mRNA was produced with a 5′ cap 1 structure (cap1) to improve translation efficiency. Free and end-stabilized mRNA were purified, buffer exchanged, and sterile filtered. The mRNAs were tested for purity and capping efficacy and were found to be >70% and >90% effective, respectively. All formulations were tested for particle size, RNA encapsulation, and endotoxin presence and were found to be <100 nm in size, with >80% encapsulation, and <10 EU/mL endotoxin.

All animal protocols and procedures were approved and conducted in full compliance with the guidelines outlined in the Guide for the Care and Use of Laboratory Animals and were approved by the University of Utah Institutional Animal Care and Use Committee (IACUC). GalT-deficient (GalT-KO) mice used in this study were constructed as previously described [21] and were not challenged with galactose since weaning. Mice were housed in standard laboratory cages within temperature (22–23 °C) and humidity (30–70%) controlled rooms under a 12:12 light–dark cycle. All mice were confirmed by molecular genotyping using previously published protocol [21]. Both male and female animals were used for the present study and were uniformly distributed among each experiment group. All WT and GalT-KO animals were weaned at 3 weeks of age, at which point they were enrolled in the experiment.

GalT-KO mice aged 3 and 6 weeks old were administered 100 µL of LNPs containing GALT mRNA via intravenous injection. The LNPs encapsulated mRNA were supplied by Moderna, Inc. (Cambridge, MA, USA). Each experimental group consisted of 8 mice (4 males and 4 females), which were randomly enrolled into treatment groups at time of weaning (Figure 2). The mice received five biweekly doses of GALT mRNA at a concentration of 2 mg/kg. No placebo or vehicle doses were administered to the wild type (WT) control or untreated GalT-KO groups for this study.

Mice underwent rotarod testing 3 and 9 weeks following the dosing regimen. The 6-lane rotarod and accompanying software (Conductor Science SA102) used in this study was purchased from Maze Engineers (Skokie, IL, USA). Mice first completed a training session on the rotarod where they would walk at a constant speed of 4 rotations per minute (rpm) for at least 180 s before starting the first trial. Three trials were conducted where the rotarod would increase from 4 to 40 rpm at a constant acceleration of 7.2 rotations per minute squared (rpm²). Maximum speed would be reached at 300 s, which was also the cutoff time for each trial. Mice would remain on the stationary rotarod between trials, where they were allowed a 5 min rest period before beginning the next trial. The trial would automatically end when IR floor sensors detected the animal to have fallen off the rod, at which point the latency would be recorded. Test investigators would manually end the trial when a mouse was observed to complete two passive rotations, classified by the mouse holding onto the rod as it makes a full rotation without walking. To combat small sample sizes, the latency from all three trials were used to compare mRNA-treated GalT-KO results to WT control and untreated GalT-KO performances. The WT control median latency was used to categorize "under-performing" and "over-performing" animals within each treatment group. Chi-squared analysis was then utilized to compare the performance distribution, or the number of animals in each group to fall into these two categories. Statistical significance was achieved when the number of over- and under-performing animals in either GalT-KO group did not follow the expected distribution demonstrated by WT control animals. Results from chi-squared tests were verified using generalized linear models (GLMs) and post hoc assessments run in R 4.4.2 software.

The composite phenotype scoring test was conducted in accordance with the protocol demonstrated by Guyenet et al. (Figure 3) [22]. This assessment consists of four parts—the ledge test, hindlimb clasping, kyphosis, and gait tests—each measured on a 0–3 scale. The overall composite score is the combined scores from each of the four assessments, where a higher score signifies a more severe manifestation of disease phenotype (score 0–12). Two blinded investigators were assigned to evaluate the composite score test where they would average the two scores given if they differed. Half-scores were granted when the behavior did not align with the full-digit score.

The ledge test requires only the home cage of the mouse and is a simple assessment of motor coordination. Each mouse was gently placed on the ledge of the cage so that its hindlimbs can grasp onto the surface. If the mouse has no obvious issues with balance and can fully support its body while walking along the ledge, it receives a score of 0. If the mouse is observed to have several missteps as it walks on the ledge but is otherwise able to maintain its balance, it receives a score of 1. A score of 2 is granted if the hindlimbs are unable to support the body and instead the mouse is dragging along the ledge using only its front limbs. Alternately, if the mouse can support its body with several missteps but demonstrates a hard landing, it is scored as 2. A hard landing is defined by the animal falling onto its head instead of its front paws when lowering itself back into the home cage. If it is incapable of walking along the ledge despite all encouragement or continues to fall off without taking a full step, this will be scored a 3.

The hindlimb clasping test is conducted by lifting the mouse from the base of its tail approximately 5–10 cm off the ground and observing the position of the hind legs. If the hindlimbs remain splayed outward from the abdomen, the mouse will be given a score of 0. A score of 1 is classified by the mouse retracting one of its legs in towards the abdomen for more than 50% of suspension time. If both hindlimbs are partially retracted towards the abdomen, it will be given a score of 2. Lastly, if both legs are completely retracted toward the abdomen, this will result in a score of 3.

The gait test serves to evaluate motor function and coordination. Mice are placed on the flat surface of a biosafety cabinet (NUAIRE Biological Safety Cabinets, Plymouth, MN, USA) and observed by investigators. If an animal fully supports itself and evenly distributes its weight while walking, it will receive a score of 0. If there is a slight limp or abnormal stride noted, a score of 1 is granted. If the mouse is seen to have its pelvis tilted inward while walking, or demonstrates "duck feet", it obtains a score of 2. The most severe score is granted when the pelvis is severely tilted inward, causing the animal to drag along the surface while walking, or a combination of less severe abnormalities (duck feet, limp, or uneven weight distribution).

Kyphosis is classified as the loss of spinal muscle tone and is a common symptom of neurodegeneration [22]. Mice are observed in both a sitting and walking position to assess kyphosis. If mild or no kyphosis is observed while sitting and the spine can fully straighten as the mouse walks, it will receive a score of 0. If mild kyphosis is exhibited while sitting but the mouse is still able to straighten its spine while walking, it will be scored a 1. If the mouse maintains mild kyphosis as it walks it will be given a 2, and if severe kyphosis is observed while sitting and walking, it scores a 3. R software was utilized to assess any changes in composite scoring behavior between treatment groups.

Data are expressed as means ± standard error of the mean (SEM). Microsoft Excel, GraphPad Prism 9 (GraphPad Software, Inc., Boston, MA, USA) and R 4.4.2 software were used to analyze data. Chi-squared analysis was utilized to determine the differences in latency distribution in rotarod performances and a Mann–Whitney U test was used for composite phenotype scoring analysis between groups. R software was used to corroborate significant findings from these analyses.

To address if experimental GALT mRNA therapy improves motor-related phenotypes in a mouse model of classic galactosemia (question 1), we compared the rotarod and composite phenotype scoring results from assessment 1 between the WT, treated, and untreated GalT-KO animals in cohort 1 (n = 8; 4 males and 4 females). Upon rotarod testing, untreated WT control mice demonstrated a median latency of 106.2 s (Table 1a). WT control median latency was used to define the performances of other groups, where any treated or untreated GalT-KO mouse with a latency less than the WT median was labeled as "under-performing". Similarly, animals that demonstrated greater latencies than the WT median values were classified as "over-performing".

We then compared the number of animals that were categorized as "under-performing" and "over-performing" across treatment groups using chi-squared analysis. In assessment 1, 85.7% of untreated GalT-KO mice performed under the WT median latency, compared to only 25% of mRNA-treated GalT-KO animals (Figure 4a). When we examined the distribution of untreated and mRNA-treated GalT-KO animals based on the WT control median values, we saw a significant improvement in the treated animals (p = 0.00005). Untreated mutant mice showed a drastically different distribution compared to the WT control animals (p = 0.001). Treatment of the GalT-KO animals proved to be effective in diminishing the genotypic difference. A GLM with a Gaussian distribution was run using treated GalT-KO mice as the reference group to confirm results obtained by chi-squared testing. This model confirmed our previous findings by showing that untreated GalT-KO animals are estimated to perform 49.37 s worse than the treated GalT-KO group (p = 0.0002) (Figure S1). Furthermore, post hoc pairwise comparison further validated these results with the untreated GalT-KO animals having a significantly lower latency than both WT control and treated GalT-KO groups (p = 0.0005 and p = 0.013, respectively) (Table S1a,b). Body weight and sex were not found to be influential factors on latency in this assessment.

We continued to see significant distinctions between the performances of the untreated WT and GalT-KO during the composite phenotype scoring test. Untreated WT control mice had a median combined score of 1, and untreated GalT-KO mice scored significantly higher (more disease phenotype) with a median score of 2.13 (p = 0.046) (Table 1b). mRNA-treated GalT-KO mice showed improvement and scored substantially lower than the untreated GalT-KO cohort with a median score of 0.88 (p = 0.05) (Figure 4b). Besides showing better performance than the untreated mutant group, the mRNA treatment of GalT-KO mice also demonstrated a return to the expected performance of WT animals (p = 0.71). To validate these findings, R software was used to run a model assessing the effect of treatment and gender on composite score performance. This model confirmed that untreated GalT-KO animals performed significantly worse than the treated GalT-KO group (p = 0.003), and that sex of the untreated GalT-KO animals was an influential factor on outcome (p = 0.0344) (Table S1c). Post hoc analysis confirmed the significant improvements in mRNA-treated GalT-KO female animals compared to their untreated counterparts (p = 0.0344) (Table S1d). Considering the rotarod and composite phenotype scoring test results, we concede that mRNA treatment of 3-week-old GalT-KO mice does improve their motor function.

To see if the positive changes resulting from GALT mRNA treatment are sustainable (question 2), we re-tested these mice after 9 weeks, or when they reached 23 weeks of age (Figure 2). For rotarod testing, untreated WT control mice had a median latency of 130 s, while untreated GalT-KO mice had a median latency of 95.5 s (Table 1a). We saw 71.4% of the untreated mutant mice performing under the WT median threshold, demonstrating a significant genotypic difference in performance distribution (p = 0.049) (Figure 5a). Treated GalT-KO mice had a median latency of 137.5 s and showed significant difference in their distribution when compared to the untreated GalT-KO cohort (p = 0.023) (Figure S2). While the untreated mutant mice mostly performed under the WT median latency, we found that only 37.5% of the mRNA-treated GalT-KO group fell below this threshold (Figure 5a). No difference was noted between WT- and mRNA-treated mutant group distribution (Figure 5a). A GLM with Gaussian distribution confirmed treatment type to have significant influence on latency. Untreated GalT-KO animals were estimated to perform 14.44 s worse than WT controls and 32.41 s worse than treated GalT-KO animals (Figure S2). This model also substantiated results from chi-squared analysis, showing that treated GalT-KO group still demonstrated significantly higher latencies than their untreated counterparts, 12 weeks following the end of their treatment (p = 0.011) (Table S2a). Again, post hoc testing in R confirmed this distinction. Though a significantly better performance was still observed in the treated GalT-KO group at this timepoint, the magnitude of effect is notably smaller than obtained in assessment 1. Comparing the two assessments, there was no apparent decline in the mRNA-treated GalT-KO group performance. Instead, we saw a trend of increased latency in the untreated mutant group over time (Figures S1 and S2). With these results, we infer that the benefits from mRNA treatment of GalT-KO animals may still exist at this point, though the impact is surely decreasing.

In the composite score phenotype scoring test, untreated WT mice demonstrated a score near to their first assessment, with a median total score of 0.875 (Table 1b). Untreated and mRNA-treated GalT-KO groups performed similarly with median composite scores of 1.25 and 1.125, respectively (Table 1b). Compared to assessment 1, both WT and untreated GalT-KO groups improved their scores, while the treated mutant group showed a slight decline in performance (Figure 5b). There were no distinctions between any of the three groups in this assessment, as confirmed with the Mann–Whitney U test and models run in R (Table S2c). Consequently, we cannot confidently confirm the benefits of mRNA treatment in GalT-KO animals have sustained over time for this test.

Finally, we compared the above results with separate groups of mice that we began dosing at an older age to address question 3 (6 weeks of age) (Figure 2). For rotarod testing, untreated WT control mice had a median latency of 126.9 s, while untreated GalT-KO mice had a median latency of 101.9 s (Table 1a). Despite the mRNA-treated GalT-KO group showing a slightly higher median latency (122.6 s) than their untreated counterparts (Figure S3), both treated and untreated GalT-KO groups followed the same performance distribution relative to the WT threshold latency (Table 1a, Figure 6a). We could not differentiate these groups as 66.7% of each population performed under the WT median latency (Figure 6a). A GLM confirmed the lack of influence that mRNA treatment had on latency in this assessment (Table S3a). While the treatment of 6-week-old GalT-KO animals did not show any improvements in rotarod performance, we observed that the untreated GalT-KO animals performed much better in the 6-week versus 3-week cohort.

For the composite phenotype scoring test, untreated WT control mice had a median total score of 0.625, and untreated GalT-KO mice had a significantly higher median score of 1.875 (p = 0.021) (Table 1b). mRNA-treated GalT-KO mice performed similarly, with a median total score of 1.5 (Table 1b). Unlike the untreated mutant group, however, the treated GalT-KO animals showed no significant deviation from WT median scores (Figure 6b). Neither treatment group nor gender elicited a significant influence on composite scoring abilities, as confirmed by GLMs run in R (Table S3).