Diet induced obesity in rats reduces NHE3 and Na(+) /K(+) -ATPase expression in the kidney.

The consumption of a high fat diet (HFD) is associated with proteinuria and altered sodium handling and excretion, which can lead to kidney disease. In the proximal tubule, the Na(+) /H(+) Exchanger 3 (NHE3) is responsible for normal protein reabsorption and the reabsorption of approximately 70% of the renal sodium load. It is the Na(+) /K(+) -ATPase that provides the driving force for the reabsorption of sodium and its exit across the basolateral membrane. This study investigates the effects that consumption of a HFD for 12 weeks has on NHE3 and Na(+) /K(+) -ATPase expression in the kidney. Western blot analysis identified a significant reduction in NHE3 and its modulator, phosphorylated protein kinase B, in renal lysate from obese rats. In the obese rats, a reduction in NHE3 expression in the proximal tubule may impact on the acidification of endosomes which are responsible for albumin uptake, suggesting a key role for the exchanger in protein endocytosis in obesity. Western blot analysis identified a reduction in Na(+) /K(+) -ATPase which could also potentially impact on albumin uptake and sodium reabsorption. This study demonstrates that consumption of a HFD for 12 weeks reduces renal NHE3 and Na(+) /K(+) -ATPase expression, an effect that may contribute to the albuminuria associated with obesity. Furthermore the reduction in these transporters is not likely to contribute to the reduced sodium excretion in obesity. These data highlight a potential link between NHE3 and Na(+) /K(+) -ATPase in the pathophysiological changes in renal protein handling observed in obesity.

Understanding STEM academics' responses and resilience to educational reform of academic roles in higher education.

BACKGROUND: Across the globe, there have been significant reforms to improve STEM education at all levels. A significant part of this has been teacher reform. While the responses and resilience of STEM teachers to educational reforms in secondary education have received significant attention, the responses and resilience of STEM teachers in higher education remains understudied. In higher education, educational reforms of academic roles have seen increasing numbers of STEM academics focussed on education. Responses of STEM academics to education reform of the academic role have some parallels with teacher resilience, but there are also potential misalignments within a culture which values and prioritises science disciplinary research. This study examined the responses of STEM academics in higher education to educational reform of the academic role using the theoretical construct of resilience and Bronfenbrenner's socio-ecological model. This was a 2-year case study of 32 academics and senior educational leaders in higher education in STEM. Data collection included semi-structured interviews which were theme coded and inductively analysed. RESULTS: The responses and resilience of STEM academics focussed on education appeared to be dependent on interactions between individual disposition in the microsystem and influences of the exosystem and the external macrosystem. Five major themes emerged about the value and quality, scholarship and expertise, progress and mobility, status and identity and community and culture of STEM academics focussed on education. The exosystem was a significant unidirectional influence on STEM academics where judgements were made concerning academic performance, awards, and promotion. Responses of senior leaders in the exosystem were influenced by the macrosystem and culture of science. Academics focussed on research, rather than education were more valued and more likely to be both financially rewarded and promoted. CONCLUSION: During this pressured decade, where COVID-19 has intensified stress, more attention on the direction and reciprocal relationships in the socio-ecological model of higher education is needed in order for educational reform in higher education STEM to be effective. Resilience of STEM academics to educational reform in higher education is a dynamic quality, and the capacity to "bounce back", learn from challenges, and realise expectations of educational reform will depend on an understanding of resilience and support of Bronfenbrenner's spheres of influence.

NEDD4-2 as a potential candidate susceptibility gene for epileptic photosensitivity.

Photosensitive seizures occur most commonly in childhood and adolescence, usually as a manifestation of complex idiopathic generalized epilepsies (IGEs). Molecular mechanisms underlying this condition are yet to be determined because no susceptibility genes have been identified. The NEDD4-2 (Neuronally Expressed Developmentally Downregulated 4) gene encodes a ubiquitin protein ligase proposed to regulate cell surface levels of several ion channels, receptors and transporters involved in regulating neuronal excitability, including voltage-gated sodium channels (VGSCs), the most clinically relevant of the epilepsy genes. The regulation of NEDD4-2 in vivo involves complex interactions with accessory proteins in a cell type specific manner. We screened NEDD4-2 for mutations in a cohort of 253 families with IGEs. We identified three NEDD4-2 missense changes in highly conserved residues; S233L, E271A and H515P in families with photosensitive generalized epilepsy. The NEDD4-2 variants were as effective as wild-type NEDD4-2 in downregulating the VGSC subtype Na(v)1.2 when assessed in the Xenopus oocyte heterologous expression system showing that the direct interaction with the ion channel was not altered by these variants. These data raise the possibility that photosensitive epilepsy may arise from defective interaction of NEDD4-2 with as yet unidentified accessory or target proteins.

Structure-function relations of biotin derivatives of apamin.

Apamin was biotinylated at various residues so as to produce an apamin derivative that was suitable for labelling apamin sensitive K+ channels. We labelled the sole histidine residue (His-18) of apamin with diazobenzoyl biocytin (DBB) and Lys-4 with NHS-biotin and NHS-XX-biotin. We found that at least two labelled species were produced by DBB. Proton NMR spectroscopy revealed that in addition to labelling the His-18, DBB labelled the Gln-16 and Gln-17 of apamin. Both NHS-biotin and NHS-XX-biotin appeared to specifically label Lys-4. To test the potency of these apamin derivatives, we developed an assay using apamin reversal of the adrenaline induced relaxation of the mouse ascending colon. The biological activity of the His-18 derivative was 46-fold less than that of native apamin. Biotinylation of Lys-4 with NHS-biotin reduced the activity by only 6-fold. The inclusion of a 14-carbon spacer between the Lys-4 and the biotin resulted in a derivative with only a 4-fold reduction in potency.

Insulin secretagogues, but not glucose, stimulate an increase in [Ca2+]i in the fetal rat beta-cell.

Fetal pancreatic islets release insulin poorly in response to glucose; however, the cellular mechanism for this is controversial. By using fura 2 to measure changes in the cytoplasmic free Ca2+ concentration ([Ca2+]i) in beta-cells, we have examined islets from fetal, neonatal, and adult rats to determine the ability of glucose and other secretagogues to cause an increase in [Ca2+]i. The effects of glucose (20 mmol/l), glyceraldehyde (20 mmol/l), leucine (20 mmol/l), arginine (20 mmol/l), and the channel effectors glipizide (50 mumol/l), BAY K8644 (2 mumol/l), diazoxide (300 mumol/l), and verapamil (20 mumol/l) on changes in [Ca2+]i were studied. In both the fetal and the mature islet, glyceraldehyde, leucine, arginine, glipizide, and BAY K8644 caused an increase in [Ca2+]i. In mature islets, glucose also increased [Ca2+]i; however, in the fetal islet, glucose had no effect on [Ca2+]i. The stimulus-induced increases in [Ca2+]i in fetal and adult islets were both significantly inhibited by the addition of either diazoxide or verapamil. Similar results were obtained when insulin secretion was measured. Our data show that various secretagogues are able to stimulate fetal islets and cause an increase in [Ca2+]i. Glucose, however, fails to cause an increase in [Ca2+]i in the fetal islet. Hence, the immature insulin secretory response to glucose by the fetal islet is due to the inability of the fetal beta-cell to translate glucose stimulation into the increase in [Ca2+]i required for exocytosis of the insulin granule.

Diphenylamine-2-carboxylate (DPC) reduces calcium influx in a mouse mandibular cell line (ST885).

Non-selective cation channels are found in many diverse cell types and have been proposed as a potential entry path for Ca2+. ST885 cells contain large numbers of these channels which are active in the resting cell. We have used Fura-2 to monitor changes in intracellular free Ca2+ ([Ca2+]i) in response to step changes in extracellular Ca2+ ([Ca2+]o). We found that DPC, a blocker of the non-selective cation channel in these cells, caused a reduction of approximately 50% in the rate of rise in [Ca2+]i following a step increase in [Ca2+]o. Since our experiments demonstrate that this phenomenon is not due to DPC blockade of Cl- channels, the Na+/Ca2+ exchanger or cyclooxygenase, we conclude that it is attributable to a direct effect of DPC on the non-selective cation channel. It thus appears that the non-selective cation channel is a significant pathway for basal Ca2+ entry in these cells.

Control of intracellular Ca2+ by adrenergic and muscarinic agonists in mouse mandibular ducts and end-pieces.

The changes in free Ca2+ ([Ca2+]i) in the cells of the secretory end-pieces and intralobular ducts of mouse mandibular glands exposed to adrenergic or cholinergic agonists were measured using fluorescence imaging techniques. [Ca2+]i in both cell types increased in a dose-dependent manner during both adrenergic and cholinergic stimulation. The duct cells responded to noradrenaline and to acetylcholine over the same concentration range (30 nmol/l to 3 mumol/l) although the maximum increase in [Ca2+]i above resting levels evoked by noradrenaline (ca. 137 nmol/l) was about twice that evoked by acetylcholine. The response to acetylcholine was blocked by atropine (0.1 mumol/l) and the response to noradrenaline was blocked by the alpha 1-adrenergic antagonist, prazosin (0.1 mumol/l), but not by the alpha 2-adrenergic antagonist, yohimbine. The alpha-adrenergic agonist, phenylephrine, mimicked the action of noradrenaline but the beta-adrenergic agonist, isoproterenol, had no effect. In contrast to the duct cells, the end-piece cells responded to acetylcholine at much lower concentrations (threshold << 1 nmol/l) than to noradrenaline (threshold ca. 300 nmol/l) and the size of the increase in [Ca2+]i above resting levels evoked by acetylcholine (216 nmol/l) was nearly 5-times greater than for noradrenaline. VIP and substance P failed to evoked a Ca2+ response in either end-piece or duct cells.

Use of replication deficient adenoviruses to investigate the role of G proteins in Ca2+ signalling in epithelial cells.

Here we report on the feasibility of using replication deficient adenoviruses to modify signal transduction systems in epithelia. We constructed two viruses, one expressing a dominant negative mutant of the alpha-subunit of Gq (Ad-EF1-dnG alpha q) and the other expressing the wild-type alpha-subunit of Gq (Ad-EF1-wtG alpha q). We used an adenovirus expressing green fluorescent protein (Ad-EF1-GFP20) to show that infection of cultured cells with an adenovirus results in at least 95% expression of the transgene in both HSG and HT29 cells. We also used an adenovirus that expresses no transgene (Ad-MX17) to demonstrate that adenoviral infection itself does not affect the resting concentration of cytosolic Ca2+ ([Ca2+]i) or the carbachol responses in these cells. We further show that Ad-EF1-dnG alpha q inhibits the increase in [Ca2+]i produced by muscarinic receptor activation in both the cell lines we studied. This inhibitory effect is not shared by Ad-EF1-wtG alpha q, which indicates that in both HSG and HT29 cells, the increase in [Ca2+]i produced by muscarinic receptor activation is largely mediated by activation of Gq. Neither virus affected the resting level of [Ca2+]i in these cells. Our findings confirm the feasibility of using replication deficient adenoviruses expressing dominant negative mutants to investigate the role of G proteins in signal transduction systems.

An inwardly rectifying K+ channel in human adenomatous parathyroid cells.

The ion channel populations of cells prepared from human adenomatous parathyroid glands have been investigated using standard patch-clamp techniques with pipettes containing isotonic K(+)-rich solutions. The principal channel type observed was an inwardly rectifying K+ channel of 35 pS conductance. In addition, we have frequently observed a lower conductance (12 pS) K+ channel that appeared to conduct current in both directions. In addition to these K+ channels, we have observed non-selective cation channels and possibly CI- channels. Although we have observed occasional current transitions that might have arisen from a large conductance K+ channel similar to that observed in rat and bovine parathyroid cells, we have not been able positively to identify such a channel in human parathyroid cells, either in cell-attached or in excised inside-out patches.

Purinergic responses in HT29 colonic epithelial cells are mediated by G protein alpha -subunits.

Using Fura-2 to measure changes in intracellular calcium ([Ca(2+)](i)), we show that P(2U)receptors in HT29 cells trigger an increase in [Ca(2+)](i)by pertussis toxin-insensitive G proteins. We then use replication-deficient adenoviruses expressing wild-type and dominant negative mutants of G(alpha q)and G(alpha i2), antisense directed against G(alpha q)or the C-terminal fragment of beta-adrenergic receptor kinase (beta ARK-CT) to identify these G proteins. We find the [Ca(2+)](i)response to UTP is not affected by increased expression of the wild-type G(alpha q), wild-type G(alpha i2)or beta ARK-CT, while it is blocked by over-expression of dominant negative G(alpha q). The timecourse of the UTP response is, however, altered by wild-type G(alpha q)and is only weakly inhibited by antisense G(alpha q). This suggests that the P(2U)response is mediated, at least partially, by a G protein distinct from G(alpha q). In contrast, the M(3)muscarinic response is inhibited by over-expression of antisense against G(alpha q), or over-expression of beta ARK-CT, a finding in agreement with our previous observation that the muscarinic response in HT29 cells is mediated by the beta gamma-subunits of G(q). We also find that P(2U)and M(3)receptors do not control identical Ca(2+)stores, suggesting that differential activation of G proteins can lead to Ca(2+)release from distinct stores.

The insulin-like growth factor-II (IGF II) receptor from rat brain is of lower apparent molecular weight than the IGF II receptor from rat liver.

The binding subunits of the insulin and insulin-like growth factor-I (IGF I) receptors from rat brain are of lower molecular weight than the corresponding receptor in rat liver, possibly due to variations in sialic acid content. We have compared the IGF II receptor from rat brain and rat liver. The brain receptor is of smaller apparent mol wt (about 10 K) on sodium dodecyl sulfate polyacrylamide gel electrophoresis. This size difference is independent of ligand binding as it persists in iodinated and specifically immunoprecipitated receptors. From studies of wheat germ agglutinin binding and the effect of neuraminidase on receptor mobility, we conclude that this difference is not simply due to variations in sialic acid content. Treatment with endoglycosidase F results in reduction in the molecular size of both liver and brain receptors and after this treatment the aglycoreceptors are of similar size. We conclude that in rat brain tissue the IGF II receptor like the binding subunits of the insulin and IGF I receptors is of lower molecular size than the corresponding receptors in rat liver. This difference is due to differences in N-linked glycosylation.

A comparison of the insulin and insulin-like growth factor I receptors from rat brain and liver.

Recent studies of the human brain insulin receptor raise the possibility that some of the differences described between rat brain insulin receptors and rat liver insulin receptors may have been due to the presence of insulin-like growth factor I (IGF-I) receptors in the brain. This study details side by side comparison of the alpha-subunits from insulin and IGF-I receptors in rat brain and liver. Insulin and IGF-I receptors are present in rat liver and brain, although IGF-I receptors are at a very low concentration in rat liver. These insulin and IGF-I receptors are clearly separate entities based on relative ligand affinities, antigenicity, and apparent molecular size. The alpha-subunits of the brain receptor for both insulin and IGF-I are about 10K lower in molecular size than the corresponding alpha-subunits from liver. These brain alpha-subunits are less sensitive to neuraminidase digestion than the corresponding liver alpha-subunit, although they do bind to wheat germ agglutinin. After treatment with endoglycosidase-F the deglycosylated apoproteins for both receptors from liver and brain have similar mol wt, suggesting that the observed differences were due to differences in N-linked glycosylation. The significance of differences in N-linked glycosylation between the brain receptors for insulin and IGF-I and the corresponding receptors in liver is not known.

Na(+)-H+ exchange in sheep parotid endpieces. Apparent insensitivity to amiloride.

We have used microspectrofluorimetry with the pH-sensitive dye, BCECF, to examine the control of intracellular pH in the secretory endpieces of the sheep parotid gland. Unstimulated endpieces in HCO3(-)-free media have a cytosolic pH of 7.5 +/- 0.03 (n = 69) which is maintained by a Na(+)-dependent proton extrusion process that can be partially supported by Li+ but not by Cs+, and is not affected by changes in extracellular Cl-, HCO3- or K+. It is not blocked by SITS or DIDS, which inhibit Na(+)-(n)HCO3- co-transport and CL(-)-HCO3- exchange, nor is it sensitive to the amiloride analogs, MIA and EIPA, which inhibit Na(+)-H+ exchangers, although very high concentrations of amiloride itself (1 mmol/l) have a (probably non-specific) inhibitory effect. It seems likely that sheep parotid secretory endpieces do contain a Na(+)-H+ exchanger that drives secretion of a HCO3(-)-rich juice, and that its insensitivity to amiloride and its analogs explains why these drugs do not block fluid secretion by the intact sheep parotid gland.

Intracellular Ca2+ release by flufenamic acid and other blockers of the non-selective cation channel.

We report in this paper using measurement of intracellular free Ca2+ with fura-2, that flufenamic acid and several related blockers of the 25 pS Ca(2+)-activated non-selective cation channel cause release of Ca2+ from an intracellular store other than the endoplasmic reticulum, possibly from mitochondria. A new compound, 4'-methyl-DPC, is found to be as effective in blocking non-selective cation channels as other flufenamate analogs but, like the parent compound, the non-selective cation channel blocker DPC, it does not cause release of Ca2+ from intracellular stores. DPC and 4'-methyl-DPC are thus the most suitable of the available blockers of non-selective cation channels for use in studies on the role of these channels in normal cell function.

Feedback inhibition of epithelial Na(+) channels in Xenopus oocytes does not require G(0) or G(i2) proteins.

Regulation of amiloride-sensitive epithelial Na(+) channels (ENaC) is a prerequisite for coordination of electrolyte transport in epithelia. Downregulation of Na(+) conductance occurs when the intracellular Na(+) concentration is increased during reabsorption of electrolytes, known as feedback inhibition. Recent studies have demonstrated the involvement of alphaG(0) and alphaG(i2) proteins in the feedback control of ENaC in mouse salivary duct cells. In this report, we demonstrate that Na(+) feedback inhibition is also present in Xenopus oocytes after expression of rat alpha,beta, gamma-ENaC. Interfering with intracellular alphaG(0) or alphaG(i2) signaling by coexpression of either constitutively active alphaG(0)/alphaG(i2) or dominant negative alphaG(0)/alphaG(i2) and by coinjecting sense or antisense oligonucleotides for alphaG(0) had no impact on Na(+) feedback. Moreover, no evidence for involvement of the intracellular G protein cascade was found in experiments in which a regulator of G protein signaling (RGS3) or beta-adrenergic receptor kinase (betaARK) was coexpressed together with alpha,beta, gamma-ENaC. Although some experiments suggest the presence of an intracellular Na(+) receptor, we may conclude that Na(+) feedback in Xenopus oocytes is different from that described for salivary duct cells in that it does not require G protein signaling.

P2X(1) receptor membrane redistribution and down-regulation visualized by using receptor-coupled green fluorescent protein chimeras.

The P2X(1) purinergic receptor subtype occurs on smooth muscle cells of the vas deferens and urinary bladder where it is localized in two different size receptor clusters, with the larger beneath autonomic nerve terminal varicosities. We have sought to determine whether these synaptic-size clusters only form in the presence of varicosities and whether they are labile when exposed to agonists. P2X(1) and a chimera of P2X(1) and green fluorescent protein (GFP) were delivered into cells using microinjection, transient transfection or infection with a replication-deficient adenovirus. The P2X(1)-GFP chimera was used to study the time course of P2X(1) receptor clustering in plasma membranes and the internalization of the receptor following prolonged exposure to ATP. Both P2X(1) and P2X(1)-GFP clustered in the plasma membranes of Xenopus oocytes, forming patches 4-6 microm in diameter. Human embryonic kidney 293 (HEK293) cells, infected with the adenovirus, possessed P2X(1) antibody-labeled regions in the membrane colocalized with GFP fluorescence. The ED(50) for the binding of alpha,beta-methylene adenosine triphosphate (alpha,beta-meATP) to the P2X(1)-GFP chimera was similar to native P2X(1) receptors. ATP-generated whole-cell currents in oocytes or HEK293 cells expressing either P2X(1) or P2X(1)-GFP were similar. Exposure of HEK293 cells to alpha, beta-meATP for 10-20 min in the presence of 5 microM monensin led to the disappearance of P2X(1)-GFP fluorescence from the surface of the cells. These observations using the P2X(1)-GFP chimera demonstrate that P2X(1) receptors spontaneously form synaptic-size clusters in the plasma membrane that are internalized on exposure to agonists.

A nucleotide receptor that mobilizes Ca2+ in the mouse submandibular salivary cell line ST885.

1. We have identified a Ca(2+)-mobilizing receptor that responds to extracellular ATP (0.1-1000 microM) in a salivary epithelial cell line (ST885). The activated receptor also stimulates Ca2+ entry from the extracellular fluid. 2. The receptor shows an agonist selectivity profile consistent with the behaviour of the nucleotide receptor class. In particular, the pyrimidine nucleotide, UTP, was equipotent with ATP. 3. Analysis of concentration-response data for the ATP and UTP-activated receptor using the Hill equation yielded EC50 values (concentrations of agonist inducing a half-maximal response) for ATP of 4.2 microM and for UTP of 4.6 microM. 4. ATP and UTP induced cross-desensitization and were not additive when administered simultaneously at maximal concentration. These findings are consistent with the hypothesis that both nucleotides act upon a common receptor. 5. The ATP analogue, 2-methylthio ATP, mobilized Ca2+ ions with higher potency (EC50 = 0.23 microM) but lower efficacy than either ATP or UTP. However, 2-methylthio ATP was not a partial agonist of the ATP/UTP receptor. At maximal concentration (30 microM), it failed to antagonize the effects of near maximal concentrations of ATP or UTP. We conclude that 2-methylthio ATP acts upon a distinct receptor.

γ-Secretase inhibition promotes fibrotic effects of albumin in proximal tubular epithelial cells.

BACKGROUND AND PURPOSE: Albuminuria is an important biomarker of renal dysfunction and is a major mediator of renal damage and fibrosis during kidney disease. The mechanisms underlying albumin-induced renal fibrosis remain unclear. There has been significant interest in γ-secretase activity in tubular epithelial cells in recent times; however, its potential role in albumin-induced fibrosis has not been investigated. EXPERIMENTAL APPROACH: The primary aim of this study was to examine the role of γ-secretase in albumin-induced fibrotic effects in proximal tubular cells. The effects of increasing albumin concentrations on fibrosis indicators and mediators in the human HK-2 cell line were examined in the presence and absence of a γ-secretase inhibitor, compound E. KEY RESULTS: Treatment with albumin resulted in a number of pro-fibrotic effects, including up-regulation of fibronectin, TGF-ß1 and the EGF-R. Interestingly, similar effects were observed in response to treatment with the γ-secretase inhibitor, compound E. Co-treatment of cells with albumin and an EGF-R inhibitor, AG-1478, resulted in significant inhibition of the observed pro-fibrotic effects, suggesting a major role for the EGF-R in albumin-induced fibrotic events. Albumin-induced effects on the EGF-R appeared to be mediated through inhibition of γ-secretase activity and were dependent on ERK-MAPK signalling. CONCLUSIONS AND IMPLICATIONS: These results provide novel insights into the mechanisms of albumin-induced fibrotic effects in tubular epithelial cells, suggesting important roles for the γ-secretase and the EGF-R. These results suggest that the proposed use of γ-secretase inhibitors as anti-fibrotic agents requires further investigation.

The interaction between megalin and ClC-5 is scaffolded by the Na⁺-H⁺ exchanger regulatory factor 2 (NHERF2) in proximal tubule cells.

Albumin endocytosis in the proximal tubule is mediated by a number of proteins, including the scavenger receptor megalin/cubilin and the PSD-95/Dlg/ZO-1 (PDZ) scaffolds NHERF1 and NHERF2. In addition, in a number of in vitro and in vivo models, the loss of ClC-5 results in a decreased cell surface expression and whole cell level of megalin, suggesting an interaction between these two proteins in vivo. We investigated if ClC-5 and megalin interact directly, and as ClC-5 binds to NHERF2, we investigated if this PDZ scaffold was required for a megalin/ClC-5 complex. GST-pulldown and immunoprecipitation experiments using rat kidney lysate demonstrated an interaction between ClC-5 and megalin, which was mediated by their C-termini. As this interaction may be controlled by a scaffold protein, we characterised any interaction between megalin and NHERF2. Immunoprecipitation experiments indicated that megalin interacts with NHERF2 in vivo, and that this interaction was via an internal NHERF binding domain in the C-terminus of megalin and PDZ2 and the C-terminus of NHERF2. Silencing NHERF2 had no effect on megalin protein levels in the whole cell or plasma membrane. Using siRNA against NHERF2, we demonstrated that NHERF2 was required to facilitate the interaction between megalin and ClC-5. Using fusion proteins, we characterised a protein complex containing ClC-5 and megalin, which is scaffolded by NHERF2, in the absence of any other proteins. Importantly, these observations are the first to describe an interaction between megalin and ClC-5, which is scaffolded by NHERF2 in proximal tubule cells.