Dynamic loading leads to increased metabolic activity and spatial redistribution of viable cell density in nucleus pulposus tissue

Abstract Background Nucleus pulposus (NP) cell density is orchestrated by an interplay between nutrient supply and metabolite accumulation. Physiological loading is essential for tissue homeostasis. However, dynamic loading is also believed to increase metabolic activity and could thereby interfere with cell density regulation and regenerative strategies. The aim of this study was to determine whether dynamic loading could reduce the NP cell density by interacting with its energy metabolism. Methods Bovine NP explants were cultured in a novel NP bioreactor with and without dynamic loading in milieus mimicking the pathophysiological or physiological NP environment. The extracellular content was evaluated biochemically and by Alcian Blue staining. Metabolic activity was determined by measuring glucose and lactate in tissue and medium supernatants. A lactate‐dehydrogenase staining was performed to determine the viable cell density (VCD) in the peripheral and core regions of the NP. Results The histological appearance and tissue composition of NP explants did not change in any of the groups. Glucose levels in the tissue reached critical values for cell survival (≤0.5 mM) in all groups. Lactate released into the medium was increased in the dynamically loaded compared to the unloaded groups. While the VCD was unchanged on Day 2 in all regions, it was significantly reduced in the dynamically loaded groups on Day 7 (p ≤ 0.01) in the NP core, which led to a gradient formation of VCD in the group with degenerated NP milieu and dynamic loading (p ≤ 0.05). Conclusion It was demonstrated that dynamic loading in a nutrient deprived environment similar to that during IVD degeneration can increase cell metabolism to the extent that it was associated with changes in cell viability leading to a new equilibrium in the NP core. This should be considered for cell injections and therapies that lead to cell proliferation for treatment of IVD degeneration.


| INTRODUCTION
Back pain is a worldwide health problem and intervertebral disc degeneration (IVDD) is frequently associated with it. 1 Cells are dependent on nutrients and waste removal to survive and function properly.
However, the cells in the IVD experience harsh conditions, especially in the central nucleus pulposus (NP). The IVD is the largest avascular organ in the human body, the nearest blood vessels to the NP core being up to 8 mm away in human lumbar IVDs. 2 Vital nutrients and toxic metabolites are transported to and from the NP cells residing within the core of the tissue mainly through the cartilaginous endplates. 3,4 While larger molecules can be transported via convective transport, 5 potentially due to the fluid flow during a diurnal cycle, 6,7 small molecules, such as glucose, lactate, and O 2 , are transported almost exclusively via diffusion. 2 Therefore, a potential onset of IVDD has been hypothesized to be due to disturbed molecular transport in the disc, 2,8 as for example the permeability of endplates decreases during IVDD. 9 Associated with this limited nutrient environment, cell density is low in the NP, with only around 2-4 million cells/ml 10,11 and with increasing disc size, the cell density decreases. 2 During IVDD, when the cells' microenvironment changes from just sufficient to deprived nutrition, cells adapt to the new conditions with an altered cell density. Spatially, this means that regions closer to the blood supply have higher densities than regions further away, where glucose demand exceeds supply. 2 In IVD organ culture under limited nutrition conditions, a new equilibrium of cell viability was reached within days. 12 The cells of the IVD adapt their energy metabolism to their microenvironment. Glucose consumption is dependent on cell density 13 and NP cells obtain their energy mainly via glycolysis, even in the presence of ambient oxygen conditions. 3,14 The metabolite lactate is extruded from cells and causes a decrease in the pericellular matrix pH. [15][16][17] This creates an imbalance of nutrient supply and waste removal that is hypothesized to lead to gradient formation where the resident NP cells experience the harshest conditions in the core of the tissue. 8 Nutrient/metabolite gradients cause metabolic rates of cells to be dependent on location in the disc. 18 Especially in the NP core, nutrient and metabolite turnover are low and critical levels for cell survival can be reached for the glucose concentration (≤0.5 mM) and pH (≤6.7) which supports only a limited cell density. 13,14,[19][20][21] Interestingly, NP cells can survive for at least 12 days without oxygen 13,14 and hypoxia does not further decrease glucose or pH induced cell death. 19 It is therefore hypothesized, that an interplay of low glucose and low pH orchestrates cellular survival, as cell survival has been reported to be pH dependent at glucose levels between 0 and 0.5 mM, indicating a synergistic effect of glucose and pH for cell survival. 2,19 Dynamic compression can increase metabolism of NP cells. 22 Therefore, it has been hypothesized that increased metabolic activity due to loading could reduce cell density 2 in a nutrient deprived environment, which could occur when attempting to regenerate the NP. However, it has not been determined if these changes in (energy) metabolism are sufficiently large enough to interact with the nutrient/ metabolite homeostasis in the NP and thus can affect cell survival.
Therefore, the aim of this study was to determine if dynamic loading leads to an increased cellular metabolic activity in the NP under a nutrient deprived environment and whether the cell viability in the NP core would be affected. This was investigated in a bovine NP explant culture platform.

| MATERIALS AND METHODS
Bovine coccygeal NP explants were cultured for up to 7 days in a chemically and mechanically controlled environment. If not stated otherwise, chemicals were purchased from Sigma Aldrich (Zwijndrecht, the Netherlands).

| Nucleus pulposus explant culture
Bovine tails from 2 to 3 year-old cows were purchased from a slaughterhouse in accordance with local regulations. NP explants were aseptically prepared as previously described 6 from the first 4-5 intact proximal discs with a 8 mm biopsy punch (4-6 mm height). The different disc levels were distributed equally and randomly among the groups. The NP explants were cultured in a previously developed volume controllable NP culture chamber 6,23 ( Figure 1A). To mimic the boundary conditions of the NP, that is, that of the inner annulus fibrosus, [glucose], oxygen tension and pH were adjusted to previously reported values in healthy and degenerated discs. 3,14,17,[24][25][26][27][28][29] Culture medium (Table S1) consisted of Dulbecco's Modified Eagle Medium mimicking either the degenerative milieu (DM, i.e., 1 mM glucose, pH 6.8) or healthy milieu (HM, i.e., 3 mM glucose, pH 7.1) with 3% fetal bovine serum (Table 1). Explants were cultured in an incubator with 5% CO 2 , 5% O 2 , and 37 C with daily medium exchanges.

| Axial compression bioreactor
As the IVD is dynamically loaded in vivo, we developed a novel forcecontrolled axial compression bioreactor to dynamically load NP explants ( Figure 1B) Figure 1C). The resulting pressure is known due to the defined surface area of the NP explants. The pressure signal is measured with load cells (Tinytronics, Eindhoven, the Netherlands), amplified and digitalized (HX711, Tinytronics) and acquired with a microcontroller (Arduino Nano Every, Arduino, Sommerville, MA, USA). Twelve explants can be loaded simultaneously with up to 1 MPa load and a maximum frequency of 1 Hz with pressure ranges of up to ±0.6 MPa per cycle.

| Loading
NP explants were equilibrated to 0.2 MPa by applying a static weight overnight. Subsequently, the volume was locked (Day 1) for the entire culture period. This confines explants to a maximal swelling of this volume while still enabling additional dynamic compression. Afterwards, explants were loaded for 6 days (to Day 7). During nondynamic loading, volume-confined samples were not additionally loaded, that is, they remained at a volume that reflects the adjusted 0.2 MPa ( Figure 1D). Simulated physiological loading consisted of 6 h nondynamic loading followed by 18 h of intermittent dynamic loading (6 cycles of: 1 h of 0.3-0.6 MPa at 0.2 Hz (square wave), and 2 h nondynamic loading). These simulated physiological loading parameters led to NP compression of 10%-20% (data not shown), which is the expected height loss of a diurnal cycle due to the fluid loss and reimbibement. 30,31 The combination of environment (healthy and degenerative milieu, HM and DM) and loading (simulated physiological and nondynamic loading, SPL and NDL) resulted in explants being cultured at one of these four conditions: HM + NDL, HM + SPL, DM + NDL, and DM + SPL ( Figure 1E); the groups were compared to NP tissue harvested on Day 0, which we will refer to as "native".

| Sample harvest
Explants were harvested on Day 0 before culture (n = 6, "native") or on Day 7 (n = 6/group). Explants were cryopreserved by plunging into liquid nitrogen to maintain the shape. The frozen samples were cut  for biochemical analysis and the remaining 1 /4 was cut into upper, core, and lower region for glucose and lactate determination in the tissue.
Additionally, on Day 2 (n = 3) explants were harvested and ½ was obtained for histology and ½ for lactate and glucose measurements in the tissue. For histological analysis, samples were covered in Tissue- Tek (Sakura, Alphen aan den Rijn, the Netherlands), frozen on dry ice, and stored at À20 C. Medium supernatant samples were taken daily before medium changes and stored at À20 C until further analysis.
Samples for biochemical as well as lactate and glucose analysis were weighed to obtain wet weight and subsequently stored at À80 C and once frozen lyophilized for 72 h, after which dry weight was determined. Dehydrated samples were digested for 16 h at 60 C in digestion buffer (100 mM phosphate buffer, 5 mM L-cystein, 5 mM EDTA) containing 140 μg/ml papain (P-5306). 32 per each region (i.e., 9 areas per sample) were counted and averaged, resulting in 3 averaged measurements per sample (top, core, and bottom). The VCD measurements in cells/mm 2 were not converted to volume in, e.g. cells/ml, to prevent conversion errors.

| Biochemical analysis
Water content of the explants was calculated from the wet weight and dry weight. Hydroxyproline (HYP) content was measured with a chloramine-T assay 34 using trans-4-hydroxyproline (H5534) as standard. Absorbance was read at 550 nm. sGAG content was determined by 1,9-dimethylmethylene blue (DMMB) binding assay at pH 3 using chondroitin sulfate from shark cartilage (C4348) as standard. 35 The absorbance at 595 nm was extracted from 540 nm. HYP and sGAG were normalized by dry weight. The fixed charge density (FCD), in mEq/g wet tissue, was calculated from the measured sGAG content (g), assuming a molecular weight of 502.5 g/mol for CS and a quantity of charge of 2 moles of charge per mole of sGAG. 36 The plate was incubated at 37 C for 30 min and absorbance was measured at 340 nm. The lactate concentration of the samples was calculated using a standard curve of sodium L-lactate (L7022) ranging from 0.016 to 1 mM.

| Statistical analysis
Statistical analysis was performed with GraphPad Prism 9.

| Dynamic loading increased lactate production
Lactate and glucose concentrations were determined in the medium samples harvested each day. Due to the higher medium glucose concentration in healthy milieu compared to degenerative milieu groups (3 mM vs. 1 mM, respectively), healthy milieu medium supernatants contained correspondingly significant more glucose than that of degenerative milieu at all time points ( Figure 3A). There was no significant difference in medium glucose concentration between the simulated physiological and nondynamic loading groups for the same medium type, however, medium glucose concentration reached critical levels for cell survival, that is, 0.5 mM, 13,14 in the degenerative milieu groups. The lactate released into the medium was significantly higher in groups loaded with simulated physiological compared to nondynamic loading groups (p ≤ 0.05, Figure 3B) in both milieus. At later timepoints, lactate release was significantly decreased compared to earlier timepoints (p ≤ 0.05). Lactate concentrations in medium of healthy milieu groups were significantly increased compared to degenerative milieu groups (p ≤ 0.05).
3.3 | Viable cell density was reduced in the core NP in dynamic loading conditions Here, we demonstrated, that dynamic loading led to increased cellular metabolic activity in ex vivo cultured NP explants as evidenced by their increased production of lactate. Furthermore, when explants were cultured in a nutrient milieu that mimics the conditions found in the inner annulus fibrosus of human degenerative discs, dynamic loading was an additional stressor that reduced the cell density in the core of NP explants, so much so that a gradient formed from the periphery to the core after 7 days. There was no gradient formation in tissue levels of glucose and lactate, indicating their quick equilibration.

| A dynamically loaded NP explant culture
The biggest obstacle for successful NP tissue cultures is its high osmotic potential. This leads to considerable swelling when unconstrained in culture and therefore needs to be counteracted. Recently, we developed a bioreactor chamber that controls swelling and enables compressive loading. 6,23 However, dynamic loading is essential for cellular homeostasis of disc cultures. 41  activity in the human and bovine lumbar disc. 46,47 Haglund et al.
reported a reduced cell viability in bovine discs after 10 days of culture at loading parameters comparable to ours but with a higher pressure range (0.1-0.6 MPa). 48 In our study, we did not observe differences in the VCD in the periphery of the NP tissue (healthy milieu with simulated physiological loading compared to native). This difference between studies could be a result of differences in the culture platform, that is, IVD organ culture versus NP tissue culture, or loading range. In our study, we did though observe a gradient in VCD going from the periphery to the center of the tissue (healthy milieu with simulated physiological loading). This spatial response is not

| Cell density reduction and nutrient/ metabolite conditions in the NP
To the best of our knowledge, this is the first study to show a reorganization of VCD in the NP core after physiological-like dynamic loading. The reduction of VCD in the NP core compared to the periphery in dynamically loaded groups additionally led to a gradient formation of VCD in the degenerative milieu. Additionally, when explants were cultured with dynamic loading, the lactate produced by glycolysis and released into the medium was increased compared to their unloaded counterparts, regardless of nutrient milieu, indicating an increased cellular metabolic activity. An increase in lactate production has also been observed in whole porcine notochordal cell-rich IVDs, when they were dynamically loaded for 1 h compared to static loaded controls. 52 As there were no changes in VCD detected on Day  Similarly, when IVD organs were cultured in a nutrient-deprived environment (2 g/L glucose, i.e., 11.1 mM), cell viability was reduced by around 50% within days, after which the cell concentration remained stable for 3 weeks. 12 Although it was clear that higher VCD gradients had formed when cells were mechanically stimulated and under more restrictive nutrient environments, we did sometimes also observe small differences in VCD between the top and bottom peripheries ( Figure 4G). This could have been due to slight differences in nutrient supply conditions between top and bottom periphery of the tissue induced by the asymmetric geometry of the bioreactor chambers. However, as these differences were within the standard deviations of the VCD of the periphery regions within in each experimental group, this was not considered significant.
The culture milieus in this study aimed to replicate the human NP environment in healthy and diseased IVDs, that is, that of the inner annulus fibrosus. Low glucose and low pH are considered to be the key parameters that control cell density in the IVD. Therefore, they were chosen as control parameters in this study. A nutrient gradient, due to diffusive transport from the vasculature to the central NP as well as consumption by the cells, is believed to cause a reduction in cell viability. 2,8 In line with this, in an artificial cartilage endplate/NP culture system, a cell viability gradient was observed, where a higher cell density had a higher glucose demand, which led to a lower cell density with increasing distance from the glucose supply. 54 While the glucose concentration remained roughly at 2.5 mM in the healthy milieu, it was near the critical value of 0.5 mM in the degenerative milieu reported to lead to NP cell death. 13,14 This means that in degenerative milieu conditions, glucose levels reached critical values even in the peripheral regions, although, as the VCD was not found to be reduced in these regions, there was enough glucose supply to support cell survival. However, there were no glucose or lactate gradients in the tissue observed at the 2 timepoints (Day 2 and Day 7).
Therefore, we expect that the differences may have been too small to detect with the harvesting and assay methods employed or that gradients were only transient, quickly resolving. While the glucose/lactate gradients can be transient, a reduction in cell density is unlikely to recover without mitogenic therapy even when nutrients are plentiful again, which can lead to reduced cell density while a nutritional gradient would be no longer measured. Nevertheless, gradients are necessary for diffusional transport, and maybe in the NP, very small, transient, gradients drive diffusional transport, even more so in critical nutrient conditions. Additionally, nutrient and metabolite gradients during IVDD might not be linear but could follow regional variations in ECM composition and cell cluster formation. 55 conditions, 3,14 and so cellular energy metabolism does not profit from such conditions. While glycolysis is less efficient, it is much faster in ATP production than oxidative phosphorylation, and the glycolysis rate of NP cells was previously estimated to be nearly 100%. 13 Additionally, NP cells express monocarboxylate transporter 4 and LDH-M, enzymes that catalyze lactate export and pyruvate to lactate conversion, respectfully. 51 The lactic acid catabolized from glucose readily dissociates to lactate and H+. Lactate can thereby either be reconverted to pyruvate via LDH or transported into the extracellular matrix. 59 If the latter occurs, the secreted lactate decreases the pH in the cells' microenvironment. [15][16][17] Interestingly, in a glucose deprived environment, lactic acidosis, that is, a low pH from a high [lactate], led to cell cycle arrest and autophagy in cancer core cells. 60,61 Similarly, this could also occur with NP cells. Once the cells are deprived of nutrients, they could react very differently and might become senescent, autophagic, apoptotic, necrotic, and so on. Nevertheless, the mechanisms of various cell fates caused by nutrient deprivation and other factors needs further investigation.

| Relevance for clinical translation and regenerative therapies
There is a controversy of whether regenerative strategies utilizing cell injections or growth factors that induce proliferation are beneficial if nutrition is limited in the harsh microenvironment in the core of the disc. 62

| Study limitations
Spatial resolution of cell density, lactate, and glucose concentration assessment was limited by the number of regions chosen, that is, the peripheral and core regions. Furthermore, the specific mechanism of cell death as a response to nutrient starvation and metabolite excess in a dynamic environment was not investigated. In this study we did not measure the pH in the medium due to the buffer system used (sodium bicarbonate/CO 2 ), which immediately changes once the medium is exposed to the environment outside the incubator. Nevertheless, the adjusted pHs might have even further decreased during culture due to the increased lactic acid production. 17 The bovine caudal discs are another study limitation, as translation to the human situation is not straightforward. NP explants were from bovine caudal discs, which are smaller than human lumbar IVDs, and smaller discs allow a higher cell density. In the human NP, the cell density is approximately 2000-4000 cells/ml, 10,11 whereas 2-3 year old bovine NPs used in this study have roughly or 9300 ± 2700 cells/ml (75 ± 22 cells/mm 2 , 8 μm slice thickness, fresh tissue), comparable to previous reports of bovine NP cellularity. 2,57,67,68 The smaller size of the bovine NP could have been responsible for a lack of a cell density gradient in the native bovine NP tissue and also for the small nonsignificant reduction in cell density in the core of the NP with nondynamic loading and healthy nutrient conditions. Nevertheless, the here presented mechanism is expected to be similar in human discs, assuming a similar health state of the cells at the baseline.
Furthermore, IVDD in humans is more complex than the here presented model. For example, a pro-inflammatory environment can additionally be involved in cell metabolism and cellular fate. 69 Therefore, future studies that mimic the degenerative disc environment should also consider investigation of the inflammatory milieu.

| Conclusion
In this study, we demonstrated that NP cells in an explant culture can survive even under very harsh conditions with critical nutrient supply, low oxygen tension, and low pH. Dynamic loading increases the metabolic activity of NP cells. When combined with a harsh nutrient environment found in degenerated discs, a new distribution with less viable cells within the NP tissue core is created. This should be considered for cell injections and therapies that lead to cell proliferation for treatment of IVD disease.

AUTHOR CONTRIBUTIONS
All authors contributed to research design and contributed to the bio-