Long-term effects of L-serine supplementation upon a mouse model of diabetic neuropathy

Abstract

Deoxysphingolipids (1-deoxySLs) are neurotoxic sphingolipids associated with obesity and diabetic neuropathy (DN) and have been linked to severity of functional peripheral neuropathies. While L-serine supplementation can reduce 1-deoxySL accumulation and improve insulin sensitivity and sensory nerve velocity, long-term outcomes have not yet been examined. To assess this, we treated 2 month old db/db mice, a model of DN, with 5–20 % oral L-serine for 6 months and longitudinally quantified the extent of functional neuropathy progression. We examined putative biomarkers of neuropathy in blood and tissue and quantified levels of small fiber neuropathy, looking for associations between lowered 1-deoxySL and phenotypes.

Toxic 1-deoxySLs were suppressed long-term in plasma and various tissue including the sciatic nerve, which is particularly targeted in DN. Functional neuropathy and sensory modalities were significantly improved in the treatment group well into advanced stages of disease. However, structural assessments revealed prominent axonal degeneration, apoptosis and Schwann cell pathology, suggesting that neuropathy was ongoing. Hyperglycemia and dyslipidemia persisted during our study, and high levels of glutathione were seen in the spinal cord. Our results demonstrate that despite significant functional improvements, L-serine does not prevent chronic degenerative changes specifically at the structural level, pointing to other processes such as oxidative damage and hyperglycemia, that persist despite 1-deoxySL reduction.

1. Introduction

The most prevalent complication of diabetes mellitus is damage to the peripheral and autonomic nervous system. Over half of diabetes patients develop diabetic neuropathy (DN), which is characterized by length-dependent sensory loss and neuropathic pain, often accompanied by other comorbidities such as autonomic dysfunction ^1–3. Metabolic stress from lipid toxicity and dyslipidemia are thought to contribute to increased risks of disease ¹. Hence the discovery of deoxysphingolipids (1-deoxySLs) in diabetic subjects has drawn attention. 1-deoxySLs is formed when the enzyme palmitoyl-transferase (SPT) uses L-alanine instead of L-serine (Fig 1). In patients as well as animal models of neuropathy, 1-deoxySL accumulation correlates with neuropathy outcomes and severity ^4,5. Unlike canonical sphingolipids, which support healthy nervous system function, 1-deoxySLs fail to undergo Golgi trafficking, are toxic to sensory neurons ^6,7, cause aberrant neuronal branching ^8–10 and impair insulin-secreting cytoskeleton in the diabetic pancreas ¹¹.

Fig 1: db/db mice have elevated plasma deoxysphingolipids and a distinct sciatic nerve 1-deoxySA:1-deoxySO profile.

(A) Biochemical synthesis of canonical (top) and neurotoxic (bottom) sphingolipids as well as their relative amount in plasma of db+/− and db/db animals. N=2 (2 animals pooled per N) for amino acid analysis and N=5 db+/− and N=5 db/db were used per condition for lipid analysis. Levels of sphingosine (SO) are consistently higher than sphingasine (SA) in the brain (B), spine (C) and sciatic nerve (D). Between db+/− and db/db, the relative levels of SA:SO is similar in spine, while db/db mice have higher levels of both lipids in brain and higher levels of SO in the sciatic nerve. Both 1-deoxySA and 1-deoxySO levels were significantly elevated exclusively in the sciatic nerve (G). In the brain (E) and spine (F), only 1-deoxySA levels were elevated. All values normalized to db/db brain SO. N=3 in db+/− condition and N=6–8 in the db/db condition, 5 – 8 months old. Welch’s t-test, *p<0.05, **p<0.01, ***p<0.001 (H). Breakdown of 1-deoxySA and 1-deoxySO by tissue compartment between db+/− and db/db, highlighting an increase in sciatic 1-deoxySO proportion.

Pathologically elevated 1-deoxySLs are found in Hereditary Sensory and Autonomic Neuropathy type 1 (HSAN1), a rare, length-dependent axonal neuropathy that shares commonalities with DN ^12,13. While genetic mutations that bias SPT towards alanine utilization contribute to 1-deoxySL accumulation in HSAN1, the role of 1-deoxySL accumulation in diabetes is not completely understood. While changes in the bioavailability of amino acids have coincided with 1-deoxySL elevation in T2DM patients ^14,15, it is unclear how and when these toxic lipids contribute to nerve damage in the long term.

We previously showed that L-serine supplementation can biochemically correct 1-deoxySL production and alleviate sensory deficits in HSAN1 mice ^16.. Importantly, Othman et al found that L-serine improved sensory nerve function in an STZ-induced rat model of diabetes after 18 weeks ⁴. To further investigate the long-term efficacy of L-serine in chronic DN, we supplemented 5–20 % oral L-serine for 6 months in 2 month old db/db mice. This model recapitulates small fiber neuropathy seen in diabetes, the earliest quantifiable measure of nerve injury in DN ¹⁷. We report on the long-term effects of L-serine supplementation upon biochemical, behavioral and pathological changes and thus shed light on timing and pathogenesis of DN.

2. Materials and Methods

2.1. Animal

We supplemented 2-month old male BKS db/db (JAX Stock Number: 000642) with the following: standard rodent chow (contains 1.2% L-serine), 5%, 10% and 20% L-serine enriched diets (Bio-Serv). Age-matched littermates (db+/−) received 1.2% or 5% L-serine. Groups of 10 db/db mice were placed on different doses of L-serine. This animal model is a well-documented type 2 diabetes and difficult-to-treat neuropathy, which has an onset time at 8 weeks after birth. At 8 weeks, db/db mice were significantly more overweight than age-matched littermates. In general, animals maintain good health until advanced disease. By 8 months of age, a subset of animals die due to diabetes-related complications.

The sample size was determined based on effects observed from the power analysis was done a priori. Animals were excluded from analysis if premature death occurs prior to study end-point. Animals were assigned by random to each treatment regimen and manually examined to ensure baseline statistics were not significantly different across groups. Supplementation was either long term (6 months) or short term (3 months), and separate cohorts were prepared, with similar sample sizes. In order to monitor the progression and severity trajectory of sensory loss (neuropathy measures) and how they interact with measures of diabetes (i.e. changes in blood glucose/weight levels), we conducted behavioral testing and blood/weight measurements monthly, over a 6-month period. Behavioral testing (for details of each test, see below sections) was done in the morning in the same experimental room. Animals were introduced to room and acclimatized for 30 min prior to the onset of the experiment. Body weight and serum glucose were recorded after a 14–16 hour overnight fast. Blood glucose was measured using a commercially available kit, AlphaTrak 2 (Abbott Laboratories) which enabled rapid serum glucose testing using tail blood and was advantageous for serial testing of the same animals across weeks without having to sacrifice the animals to draw blood. At the end of the study, animals were sacrificed at 8 months of age, at which point samples were collected for further biochemical analysis including lipid analysis, which required 1mL of total blood.

2.2. Mechanical and Thermal Sensory Testing

Mechanical sensitivity was assessed using Von Frey filaments by examiners blinded to supplementation dose. The withdrawal threshold was recorded as the lightest filament that elicited two positive responses per 10 applications of stimuli. Thermal sensitivity was tested using the hot plate, using a two-minute habituation at 37 degrees Celsius, followed by behavioral assessment of withdrawal at 50, 52 and 54 degrees Celsius. Paw withdrawal time was recorded and averaged across three trials per temperature. Cut-off time was set at 20 seconds to avoid tissue damage. Animals were tested on a monthly basis until the time of sacrifice.

2.3. Nerve Conduction Studies

Nerve conduction tests were performed in db+/−, db/db and L-serine supplemented db/db at 6 months after the study. Experiments were conducted at the Rutkove laboratory of Beth Israel Deaconess Medical Center by scientists blinded to supplementation dose. After anesthesia was applied, surface sensory and motor recording electrodes (0.7 mm sensory needles insulated with Teflon) were used to perform measurements of sensory and motor velocity and amplitudes. Prior to statistical analyses, outliers were removed by following statistical guidelines.

2.4. Gastrointestinal Transit Assay

To assess autonomic function, gastric transit assay was done based on a previous study ¹⁸. An 18-gauge feeding needle was used to administer a 200uL solution of 10% methylene blue in dextrose by oral gavage by examiners blinded to supplementation. After 15 minutes, the mice were euthanized and the small bowel was gently pulled straight. The distance between the gastroduodenal junction and the dye front was recorded as a measure of gastric motility.

2.5. Sciatic Nerve Electron Microscopy and Ultrastructure Analysis

Histology was collected at 8 months of age. The distal sciatic nerve (1 cm from the sciatic notch) was harvested without transcardial perfusion and fixed in a 2.5% glutaraldehyde solution. Nerves were then fixed in 1% osmium tetroxide and embedded in araldite resin. Cross sections of the nerve were imaged at 11000x to examine unmyelinated fibers under an electron microscope. Axon diameter was manually quantified by tracing the inner circumference to infer axon diameter using Photoshop and ImageJ.

2.6. Lipids and Blood Chemistry

Sphingoid bases in plasma and tissue were analyzed as described previously ¹³. Isotopically labeled SA (d7), SO (d7) and deoxymethyl-SA (d5) were used as internal extraction and chromatography standards (Avanti Polar Lipids). The investigator performing the lipid analysis was blinded to genotype or dietary intervention. Plasma glucose, LDH, ALT and lipase levels were analyzed by the MGH Core for Clinical Pathology.

2.7. Amino Acid and Metabolite Analysis

Plasma and tissue amino acids and metabolites were extracted and assessed at the Whitehead Institute for Biomedical Research at the Massachusetts Institute of Technology. Isotopically labeled phenylalanine-13C9–15N was used for pyruvate and lactate, and proline-13C9–15N was used for glucose detection as internal run standards. For amino acid analysis, 2 animals were pooled together with equal parts of blood per sample. Valine-13C9–15N was used as an internal run standard for serine, alanine and glycine. All values were normalized to tissue weight as an external standard. Plasma and tissue total glutathione was assayed using a commercially available kit (Enzo Life Sciences).

2.8. Blinding

Only C.X was aware of group allocations. Group allocation keys were stored in an excel spreadsheet, which was not accessible by C.X during the experiment which include blood glucose, sensory and thermal tests. Y.G, who performed oral gavage experiments, was blinded to the treatment. S.S and T.H performed lipid biochemistry experiments and were blinded to genotype and treatment. M.S performed electron microscopy experiments and was blinded to genotype and treatment. J.L and S.R performed nerve velocity experiments and were blinded to genotype and treatment. C.X performed all subsequent data analysis.

2.9. Outcome Measures

Outcomes of functional and behavioral studies were defined as follows: mechanical and heat sensitivity (latency to response, at which parameter), blood glucose (mg/dL), weight (g), sensory nerve velocity (ms), motor velocity (ms), autonomic function/gastrointestinal motility (distance travelled by dye injected through the esophagus). For biochemical studies, lipids (concentration) and amino acid (concentration) was measured normalized to an internal standard. For ultrastructural studies, outcome measures were defined as distribution and number of axons for a specific diameter range. Since we calculated sample size (animal per group) prior to the start of the experiment, we used the same sample size for functional and behavioral studies, with the exception of lipid and blood metabolite measurements, which required samples to be split between the two experiments.

2.9. Statistical Analysis

Data are shown as mean ± SEM. For time-dependent studies, two-way ANOVA was performed, followed by the Bonferroni correction for multiple comparisons. For terminal studies, either the One-way ANOVA followed by Dunnett’s correction for multiple comparisons or the Mann-Whitney t-test was performed to compare db+/− with db/db unsupplemented and db/db receiving L-serine supplementation (5%, 10% and 20%). Analyses were run on GraphPad Prism (GraphPad Software, Inc., San Diego, CA). In general, nonparametric tests were used where possible to ensure no assumptions were made on underlying distributions.

3. Results

3.1. De novo synthesis of sphingolipids in db/db mice is altered in plasma and the peripheral sciatic nerve

First, we verified that both 1-deoxysphinganine (1-deoxySA) and 1-deoxysphingosine (1-deoxySO) levels were significantly elevated in the plasma of untreated db/db mice compared to db+/− at 5–8 months of age (p<0.0001 and p<0.01), while canonical sphingolipids (sphinganine and sphingosine) remain unaltered (Fig 1A). Plasma L-serine levels were slightly decreased (although not significantly), while L-alanine levels remained unchanged. Next, we measured lipid profiles across brain, spine and sciatic nerve in db+/− and db/db mice. As DN primarily affects the peripheral nervous system, we verified that accumulation would be higher than in other tissues. In the sciatic nerve, overall 1-deoxySL was 4-fold higher than in brain and 2-fold higher than in spine. Canonical sphingolipids were significantly increased in the brain and sciatic when comparing db/db to db+/− (Fig 1B, C and D). Interestingly, while 1-deoxySA levels were increased across all compartments, 1-deoxySO was increased exclusively in the sciatic nerve (Fig 1E, F and G). Furthermore, the increased contribution of sciatic-derived 1-deoxySO to total 1-deoxySL in db/db nervous tissue suggests that a greater proportion of 1-deoxySA were converted to 1-deoxySO exclusively in the peripheral nerve (Fig 1H). This suggests that 1-deoxySO localizes to affected tissue and may have a pathological role.

3.2. L-serine supplementation suppresses global and tissue-specific 1-deoxySL production and correlates with improved gastric function

We next analyzed plasma 1-deoxySL and canonical sphingolipid levels at the endpoint of the study (6 months of supplementation) and found 1-deoxySL production reduced without significantly altering canonical lipids (Fig 2A and B). L-serine significantly suppressed 1-deoxySA production exclusively in the sciatic nerve (Fig 2H, Mann-Whitney t-test, p<0.05, Fig 2C, D and E). Reduced levels of plasma 1-deoxySA and 1-deoxySO significantly correlated with improved gastric motility (Fig 2I, K, Spearman rho correlation, r=0.77, p=0.0049 and r=0.66, p=0.0238), while 1-deoxySO better correlated with sensory nerve velocity compared to 1-deoxySA (Fig 2J, L, r=0.5, n.s and r=0.048, n.s.). Overall, L-serine supplementation replenished levels of L-serine and L-glycine in the plasma without altering L-alanine levels (Fig 2, M-O).

Fig 2: Serine supplementation suppresses 1-deoxySL accumulation in plasma and tissue and correlate with improved sensory function.

(A) Plasma deoxysphingolipids (1-deoxySA and 1-deoxySO) are significantly reduced in dose-dependent L-serine supplementation in plasma. (B) Canonical sphingolipids (SO and SA) not significantly different with L-serine. In brain (C,F), spine (D,G) and sciatic (E,H), canonical sphingolipids not significantly different. In sciatic, both 1-deoxySA and 1-deoxySO are significantly reduced. Lower plasma 1-deoxySA (I) and 1-deoxySO (K) correlated significantly with improved gastric motility; sciatic nerve conduction speed correlated with reduced 1-deoxySO but not with 1-deoxySA. Plasma L-serine levels increased in supplemented db/db. (B) L-alanine levels were not obviously altered in plasma. (C) Plasma L-glycine is decreased in db/db relative to db+/− and was restored with L-serine. (D) Spine versus sciatic amino acid profile of L-serine. (E) L-alanine levels are increased in db/db spine regardless of diet and are mildly decreased in db/db sciatic regardless of diet. (F) L-glycine levels are increased in db/db spine relative to db+/− and further increases with L-serine but remains low in db/db regardless of L-serine supplementation. N=2–3 per condition, with 2 animals pooled into each N. No significance was detected due to limited N. Plasma (M-O) and tissue (P-R) amino acid levels in db+/−, db/db and db/db L-serine. (S-U) Glucose metabolites are unchanged with L-serine supplementation. (V) Schematic summarizing metabolite levels in the spine vs sciatic of db/db and db/db serine. Metabolites marked as increased/decreased if the mean value is +/− 1 SEM from the db+/− control group.

3.3. Differences in L-serine supply explain 1-deoxySL accumulation and selective vulnerability to neuropathy

We hypothesized that if L-serine is indeed tied to pathology, low regional L-serine bioavailability should contribute to localized pathology from 1-deoxySL accumulation. Furthermore, the treatment group receiving oral L-serine should have normalized L-serine levels in affected tissue, compared to nonaffected tissue. To test this, we quantified three major amino acids metabolized by SPT (L-serine, L-alanine and L-glycine) using LC-MS, comparing spinal cord (relatively spared in DN) and sciatic nerve (target of distal neuropathy) in db/db vs. db+/− mice.

Interestingly, in the untreated sciatic nerve, L-serine was reduced by 50% in db/db (Fig 2P) compared to db+/− and in turn, supplementation restored sciatic L-serine levels, although the variance was high across animals. In contrast, spine L-serine was higher in db/db than in db+/−. Importantly, L-serine supplementation was effective in increasing L-serine levels in both sciatic nerve and spinal cord, suggesting that oral L-serine could reach both central and peripheral nervous tissue. L-serine treatment did not alter abundance of L-alanine or L-glycine (Fig 2Q-R). Metabolites such as lactate and glucose remained elevated in db/db even after L-serine supplementation (Fig 2S–U). Qualitatively, our data suggests that in the spinal cord, amino acids (L-serine, L-alanine and L-glycine) and glucose metabolites (pyruvate and lactate) were generally increased in db/db compared to db+/−, whereas in the sciatic nerve they were reduced or unchanged (Fig 2V). This suggests the metabolic content of spine tissue trends towards pro-resource, whereas the sciatic trends towards resource decline. While L-serine supplementation appears to replenish L-serine shortage in the sciatic nerve, it is unclear whether replenishing other metabolites in the peripheral nerve may further benefit treatment.

3.4. L-serine supplementation alleviates mechanical desensitization, thermal hypoalgesia and improves sensory nerve function in the sciatic nerve and enteric system

We followed a cohort of db/db animals over a 6-month period receiving varying doses of L-serine supplementation beginning at 2 months of age, which is believed to predate the onset of neuropathy (18) (Fig 3). As expected, untreated db/db mice developed mechanical desensitization (p<0.0001 at 7 months of age) and progressive loss of thermal sensitivity (Fig 3 A and B). Animals supplemented with L-serine, regardless of dose, showed reduced desensitization slopes and better prognosis throughout the disease timeline. By the 5^th month of intervention, the 5% group performed best (p<0.0001) and had a 3-fold increase in mechanical sensitivity compared to untreated db/db animals. Thermal hypoalgesia was similarly alleviated in L-serine treated db/db mice, as evidenced by a shortened withdrawal response seen on the hot plate. Higher doses of L-serine (20%) did not confer additional therapeutic benefits.

Fig 3: L-serine supplementation alleviates sensory deficits in mice independent of hyperglycemia and weight gain.

(A) Response to mechanical stimulation represented as folds increase from db+/−. Groups on L-serine exhibit improvements. (B) L-serine supplementation preserves thermal sensitivity at all doses of L-serine. (C) Fasting glucose levels are not altered significantly. (D) db/db mice regardless of diet, remain significantly more obese throughout the trial. (E) Gastrointestinal transit deficits were rescued with L-serine supplementation. (F) Impaired sensory conduction is restored to db+/− levels after L-serine. (G) Deficits in motor conduction velocity were unchanged. N=7–10 animals were analyzed per condition. ANOVA followed by Bonferroni correction for Figs A-D. Mann-Whitney t-test for Figs E-F. *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001

At the end of the supplementation period, we selected the intermediate dosage (10%) cohort for sensory nerve conduction testing of the sciatic nerve, which is commonly impaired in neuropathy (Fig 3F). Remarkably, the 10% group exhibited a sensory velocity comparable to that of db+/− controls (p<0.05), independent of changes in motor velocity, suggesting a sensory neuron-specific effect. As diabetic neuropathy also affects the enteric nervous system, we assessed gastrointestinal motility as a measure of enteric function. We found that the distance traveled by the dye front after oral gavage in the 5% and 20% group was similar to that of db+/− mice, with the 10% trending towards significance in agreement with earlier sensory testing data (Fig 3E). Together, our findings suggest that long-term L-serine supplementation is neuroprotective across sensory and autonomic functional measures.

Of note, when we looked at the effect of L-serine on common diabetic markers such as weight and blood glucose, we found no improvements after supplementation (Fig 3C and D). All db/db mice remained significantly heavier than db+/− mice at all time points. Glucose metabolites were comparable between untreated db/db and treated groups and were persistently higher than in db+/− controls (Fig 2S-U). We did not assess blood HbA1c levels but suspect similar trends as those seen in glucose, given that these values are highly correlated ¹⁹. Together, our data suggests that L-serine may act independently of glucose-mediated oxidative stress in pathology.

3.5. L-serine preserves small unmyelinated fibers in early but not advanced disease

Given the significant functional rescue in the treatment group, we were curious whether this was reflected in improvements in nerve fiber ultrastructure. To test whether L-serine halts small fiber neuropathy progression, we assessed the extent of neuropathy using electron microscopy, comparing untreated and treated db/db mice at early and advanced disease time points. We confirmed previous reports that unmyelinated nerve ultrastructure was compromised in untreated db/db mice We observed instances of axonal degeneration (appearance of swollen axons with microtubules), presence of regenerating axons (appearance of abnormally small axons), vacuoles, apoptotic mitochondria and irregular axonal spacing within the Schwann cell sheath (Fig 4B) in 5 months old db/db mice. We contrasted these results with animals treated during early disease (3 months of treatment, beginning at 2 months of age, onset of neuropathy). L-serine treated db/db mice retained twice as many axons in the 200–400 nm range, resulting in a slight leftward shift of axon diameter distribution (Fig 4A). This suggested that L-serine treatment prolonged the survival of small-sized unmyelinated fibers, likely by reducing instances of axonal degeneration. This improvement occurred independent of average number of axons (Fig 4C). Thus, L-serine supplementation appears to have neuroprotective effects during early disease, which appears also be supported by biochemical evidence showing delayed toxic 1-deoxySA and 1-deoxySO build-up in early disease (Supplementary Figure 1A-B).

Fig 4: Alleviated axonal swelling in small unmyelinated fibers following 3 months and severe degeneration after 6months of L-serine supplementation.

(A) Representative electron micrographs of db/db and db/db 10%. (B) Axonal histogram distribution of unmyelinated axons in the sciatic nerve from 3-month supplementation, showing decrease in percentage of abnormally small axons (arrow). (C) No significant difference in average axon number. (D) Severe Schwann cell pathology and fragments containing degenerating axons (arrowheads). (E) Schwann cell cytoplasm have a swollen/watery appearance and large gaps within the Remak bundle (arrowheads). (F) Altered axonal sorting where a Schwann cell no longer supports one but multiple axons, resulting in morphological alterations (asterisks). (G) Axons with intact (arrowheads) and apoptotic (asterisks) mitochondria residing in the same pathological Remak bundle. (H) Severe axonal swelling and apoptosis (asterisks), coexisting with abnormally small axons. (I) Bands of Bungner showing apoptotic and normal axons. Scale bar = 500nm. (J) Plasma glutathione is increased in untreated db/db but normalized in L-serine treated mice. (K) Spine glutathione levels were not significantly different in db/db mice treated with varying L-serine doses.

In a separate cohort, we investigated whether the short-term neuroprotective effects extended into advanced stages of diabetic neuropathy (8-months of age). To our surprise, we did not observe structural improvements in L-serine treated unmyelinated fibers compared to untreated controls. Regardless of treatment, animals exhibited persistent abnormalities in axonal sorting (Fig 4F), mitochondrial apoptosis (Fig 4G) and bands of Bungner (columns of Schwann cells that contain normal, apoptotic and regenerating axons) (Fig 4I). As a measure of global and tissue oxidative stress, we quantified total glutathione levels in plasma and spinal cord and found that while plasma glutathione of the treatment grouop was restored to control levels, tissue glutathione levels only slightly decreased with a low L-serine diet but increased in the 20% dose group (Fig 4I-J), suggesting persisting neuronal oxidative stress. These results suggest that L-serine supplementation may have short term benefits on small fiber ultrastructure but fails to prevent deteriorating fibers over time. Nevertheless, long-term L-serine treatment leads to significant improvements in functional neuropathy, marked by improved sensory nerve conduction, thermo and tactile sensitivity as well as autonomic function.

4. Discussion

We report that chronic pathology in a type 2 diabetes neuropathy model persists despite successful reduction of global and tissue-specific 1-deoxySLs by L-serine supplementation. While systematic flooding with L-serine, the canonical substrate for sphingolipid synthesis, can antagonize the production of 1-deoxySLs in DN leading to behavioral, sensory and autonomic improvements of functional neuropathy, our neuropathology demonstrates axonal swelling after 6 months of supplementation (at 8 months of age), coinciding with elevations in spinal cord glutathione levels. We speculate that L-serine was not able to rescue damage caused by 1-deoxySLs earlier in life or that metabolic stress, such as those from persistent hyperglycemia and hyperlipidemia independent of 1-deoxySLs, cannot be fully corrected by L-serine alone.

Prior transcriptomic and metabolomic studies have revealed tissue-specific changes that may help explain the heterogeneity of manifestations in diabetes ²⁰. Here we show that 1-deoxySO selectively accumulates in the sciatic nerve - a prominent site of distal neuropathy - up to 13-folds higher compared to brain. This coincides with localized L-serine deficiency, which may predispose to greater 1-deoxySLs synthesis and lead to peripheral nerve damage early in disease. Our data suggests that while L-serine supplementation was neuroprotective for small unmyelinated fibers in early stages of disease, it did not halt axonal degeneration and Schwann cell pathology in the long term, suggesting that 1-deoxySLs may have a role earlier in disease progression, as we began supplementation prior to the onset of neuropathy, but not as early as from birth. Othman et al have shown that L-serine supplementation immediately following streptozotocin injections was more effective at preserving small fiber structure than supplementation after the onset of neuropathy ⁴. Thus, it is possible that damage due to early 1-deoxySLs production cannot be reversed with later biochemical correction. Persistent, uncontrolled hyperglycemia and hyperlipidemia additionally contribute to the long-term effects of DN. In the future combination therapy targeting both 1-deoxySLs and hyperglycemia may yield synergistic benefits.

The role of L-serine and 1-deoxySLs during metabolic stress has recently gained more attention. L-serine dysregulation is implicated in rodent models of ALS ²¹, neuropathic pain in pancreatic cancer patients ²², diabetic retinopathy and macular telangiectasia type 2 (Mac Tel) ^23,24. In turn, L-serine supplementation restores insulin sensitivity in the liver ²⁵ and glucose homeostasis ²⁶. Furthermore, restriction of dietary serine and glycine has been shown to significantly suppress tumor growth due to 1-deoxySL buildup ²⁷. Our study demonstrates that L-serine deficiency localizes to the affected peripheral nerve in DN and further shows that long term treatment restores L-serine levels and keeps 1-deoxySL accumulation at bay. Thus, the prevalence of 1-deoxySL detection across multiple disorders may suggest that 1-deoxySL itself is a biomarker of “metabolic stress”. We speculate that in this scenario, the promiscuity of SPT then shifts the affected area to produce more 1-deoxySL due to L-serine shortage. Finally, buildup of 1-deoxySL leads to cell and tissue death, which is analogous to a sinking ship cutting off other accessory components to keep the essential part afloat.

5. Conclusion

In conclusion, we demonstrate that L-serine supplementation in db/db mice suppresses 1-deoxySLs but oxidative stress, hyperglycemia and neurodegeneration persist. While global and tissue-specific 1-deoxySLs remained consistently low with even our lowest L-serine regimen (5%), this did not confer long-term benefit to the recalcitrant pathology, although there is a significant functional improvement in sensory nerve conduction and autonomic function. Given that L-serine has led to clinical improvements in adults with HSAN1 and is well-tolerated ²⁸, a combination treatment regimen for DN, focusing on early intervention with L-serine, together with glycemic control and other conjunctive therapy, needs to be next investigated.