Comparison of intestinal glucose flux and electrogenic current demonstrates two absorptive pathways in pig and one in Nile tilapia and rainbow trout

Abstract

The mucosal-to-serosal flux of ¹⁴C 3-O-methyl-d-glucose was compared against the electrogenic transport of d-glucose across ex vivo intestinal segments of Nile tilapia, rainbow trout, and pig in Ussing chambers. The difference in affinities (K_m “fingerprints”) between pig flux and electrogenic transport of glucose, and the absence of this difference in tilapia and trout, suggest two absorptive pathways in the pig and one in the fish species examined. More specifically, the total mucosal-to-serosal flux revealed a super high-affinity, high-capacity (sHa/Hc) total glucose transport system in tilapia; a super high-affinity, low-capacity (sHa/Lc) total glucose transport system in trout and a low-affinity, low-capacity (La/Lc) total glucose transport system in pig. Comparatively, electrogenic glucose absorption revealed similar K_m in both fish species, with a super high-affinity, high capacity (sHa/Hc) system in tilapia; a super high-affinity/super low-capacity (sHa/sLc) system in trout; but a different K_m fingerprint in the pig, with a high-affinity, low-capacity (Ha/Lc) system. This was supported by different responses to inhibitors of sodium-dependent glucose transporters (SGLTs) and glucose transporter type 2 (GLUT2) administered on the apical side between species. More specifically, tilapia flux was inhibited by SGLT inhibitors, but not the GLUT2 inhibitor, whereas trout lacked response to inhibitors. In contrast, the pig responded to inhibition by both SGLT and GLUT2 inhibitors with a higher expression of GLUT2. Altogether, it would appear that two pathways are working together in the pig, allowing it to have continued absorption at high glucose concentrations, whereas this is not present in both tilapia and trout.

INTRODUCTION

In the 1960s, the mechanism of glucose transport across the brush-border membrane (BBM) of mammalian enterocytes was first proposed (16, 17). Since then, this model developed to include both the entry of glucose through the sodium-dependent glucose transporter type 1 (SGLT1) by secondary active transport, followed by exit across the basolateral membrane via the facilitative glucose transporter type 2 (GLUT2) (8, 16–18). However, 11 more sodium-dependent glucose transporters have been identified and characterized that may also contribute to glucose movement across the intestinal BBM in addition to SGLT1 (1, 4, 5, 9, 19, 23).

Generally, it is thought that SGLTs saturated around 30–50 mM glucose, reaching their maximum transport capacity (15–17, 21, 44). However, luminal/apical glucose concentrations in mammals are often in excess of 50 mM (thought to be around 300–400 mM), especially after a meal (15, 21). Therefore, the luminal/apical glucose concentration will exceed the maximum saturation capacity of SGLTs (15, 21, 44). Thus this leads to the question of whether SGLT transporters are the only ports of glucose entry present along the apical membrane of enterocytes, especially when luminal/apical glucose concentrations exceed 30–50 mM.

Therefore, the proposed theory of apical glucose absorption was modified to suggest the involvement of two components (15, 17, 44). This includes the SGLTs and a diffusive component that saturates at higher concentrations than SGLTs (15–17). This diffusive component has been argued to follow different pathways, including a “paracellular solvent drag” pathway and a “GLUT2 translocation” pathway (16, 17, 44). However, the evidence gathered was in stronger support of the “GLUT2 translocation pathway” that suggested GLUT2 was responsible for the majority of glucose absorption once luminal glucose concentrations exceeded the capacities of SGLTs (17, 44). This pathway was found to be initiated after SGLT1 reached maximum saturation capacity, which then activates protein kinase C βII (PKC βII), which activates downstream pathways leading to GLUT2 translocation to the BBM of enterocytes (17, 44). Once situated on the apical side, GLUT2 assists in glucose absorption at high luminal glucose concentrations (17). However, this has not been confirmed outside of the mammalian rat and obese human ileal samples or with detailed functional analysis (3, 17). Understanding the presence or absence of this putative mechanism in different species would be beneficial in providing alternative targets in controlling blood glucose levels and help elucidate appropriate models to study this mechanism.

Additionally, functional studies of apical GLUT2 involvement to assist in the transport of glucose at high luminal concentrations has been lacking in other mammalian models like the pig (Sus scrofa) and nonexistent in fish species. Therefore, pigs were used because they have a similar gastrointestinal physiology to humans and comparable nutrient requirements, more so than rodent models (1). The pig provides an interesting comparison in terms of glucose absorption with fish species, highlighting possible divergent differences between terrestrial and aquatic vertebrates. Finally, the use of Nile tilapia (Oreochromis niloticus) and Rainbow trout (Oncorhynchus mykiss) here adds to our previous work characterizing sodium-dependent glucose transport along the fish intestine, as well as validating the methods and results in this study (37).

Here, we present the mucosal-to-serosal flux (J_ms) of 3-O-methyl-d-[¹⁴C]glucose (nonmetabolizable form of d-glucose), as well as the concurrent electrogenic transport (∆I_sc) of d-glucose across the Nile tilapia proximal intestine, rainbow trout midgut segment, and porcine jejunum in Ussing chambers. The sodium-dependent transport of glucose has been characterized previously in each of these segments (36, 37). More importantly, all of these express high levels of SGLT1, where SGLT1 activation would be needed for GLUT2 translocation (17, 36, 37, 44). Therefore, these intestinal segments were chosen to be studied here. Overall, our data suggest functional GLUT2 involvement in the porcine jejunum, assisting in continued glucose absorption at high concentrations. In contrast, this mechanism of glucose absorption appears to be absent in both tilapia and trout. Altogether, supported by K_m “fingerprints,” inhibitors, and gene expression, the porcine jejunum appears to represent a different absorptive pathway of total glucose absorption in comparison to teleost species tilapia and trout, especially at high luminal glucose concentrations.

MATERIALS AND METHODS

Maintenance of Animals

All fish and pigs were maintained in accordance with the guidelines of the Canadian Council on Animal Care (CCAC, 2005) (6). All animal protocols were approved by the Animal Care Committee at the University of Saskatchewan (AUP No. 19980142 for fish and 20130034 for pig).

Pigs.

Seven- to nine-week-old purebred Yorkshire barrows were obtained from a finishing farm. They were housed in pairs in the Animal Care Unit at the University of Saskatchewan and provided with a commercially prepared, nonmedicated, pellet starter diet to satiation/ad libitum. At the time of study, the pigs were fed to satiation before euthanization. The contents of the feed contained the following: crude protein (16%), crude fat (2.4%), crude fiber (5%), vitamin A (6,000 IU/kg), vitamin D3 (900 IU/kg), vitamin E (25 IU/kg), calcium (0.62%), phosphorous (0.55%), sodium (0.2%), lysine (0.83%), methionine (0.27%), and threonine (0.58%) (Whole Earth Pig Grower, Co-op Feed).

Nile tilapia.

Nile tilapia (female) were obtained from AmeriCulture (Animas, NM) as fingerlings. They were housed in 360-liter tanks filtered via a biological filtration system. Photoperiod was kept constant at 14:10-h light-dark cycle, and the municipal, dechlorinated water temperature was maintained at 27 ± 2°C. Fish were fed a standard ration of commercial feed by hand twice per day to ~2–5% of their body weight (maintenance ration) or to visual satiation similar to ad libitum. At the time of study, the fish were fed to satiation before euthanization. The contents of the feed contained the following: crude protein (54%), crude fat (16%), crude fiber (1%), vitamin A (15,000 IU/kg), vitamin D3 (5,000 IU/kg), vitamin E (400 IU/kg), calcium (2.3%), phosphorous (1.5%), and sodium (0.2%) (EWOS). Finally, the average weight of fish used was 700 g.

Rainbow trout.

Triploid female rainbow trout were obtained from Wild West Steelhead (Lucky Lake, Saskatchewan, CA). The fish were housed in standard 1,000- to 4,000-liter tanks, provided with biologically filtered recirculation systems. They were housed in municipal, dechlorinated water at temperatures between 11 ± 2°C, with a photoperiod at 12:12-h ligh-dark cycle. They were fed a standard ration of commercial feed to ~2–5% of their body weight (maintenance ration) or to visual satiation similar to ad libitum. At the time of study, the fish were fed to satiation before euthanization. The contents of the feed contained the following: crude protein (53%), crude fat (20%), crude fiber (1%), vitamin A (15,000 IU/kg), vitamin D3 (5,000 IU/kg), vitamin E (300 IU/kg), calcium (2.3%), phosphorous (1.6%), and sodium (0.2%) (EWOS). Finally, the average weight of fish used was 300 g.

Ex Vivo Tissue Collection

Tilapia and trout.

Fish were euthanized by blunt trauma to the head, and the entire intestinal section was dissected out of both fish. In Nile tilapia, its intestinal section was separated to obtain proximal intestinal tissue (2 in. distal from the stomach). In rainbow trout, the intestine was separated to obtain the midgut section (located right after the pyloric ceca, 2 in. from the stomach).

Pig.

Pigs were euthanized by penetrative captive bolt resulting in sudden death (as per the CCAC and American Veterinary Medical Association guidelines), and the entire gastrointestinal section was removed for further dissections (28). The intestine was separated to obtain sections of the jejunum, which was obtained 18 in. distal from the stomach.

Electrophysiology

Ussing chamber technique.

The fish intestinal sections (tilapia proximal intestinal and trout midgut) were examined in Ussing chambers using techniques adapted from this group (24, 25, 32, 37, 39). The Ussing chamber system used in this study was an EasyMount Ussing Chamber System (Physiologic Instruments, San Diego, CA). Sections were washed in teleost saline containing the following (in mM): 118 NaCl, 2.9 KCl, 2.0 CaCl₂·2H₂O, 1.0 MgSO₄·7H₂O, 0.1 NaH_2.PO_4·H₂O, 2.5 Na₂HPO₄, and 1.9 NaHCO₃ at pH 7.9 (35a). This physiological buffer was used for both tilapia proximal intestine and trout midgut. No glucose was added to the buffer. The lumen of the intestine for both fish was also washed with buffer using an 18-gauge needle to clean the luminal membrane of any residual chyme.

The pig intestinal section (jejunum) was examined in Ussing chambers similarly to the fish (36). The jejunal segment was placed in Krebs buffer containing the following (in mM): 114 NaCl, 5 KCl, 2.15 CaCl₂·2H₂O, 1.1 MgCl₂·6H₂O, 0.3 NaH_2.PO_4·H₂O, 1.65 Na₂HPO₄, and 25 NaHCO₃ at pH 7.4, similar to the buffer used in a mice study done by Clarke (7). No glucose was added to the buffer. The lumen of the intestine was washed with buffer using an 18-gauge needle to clean the luminal membrane of any residual chyme. The muscularis mucosal layer was also stripped from the intestine before mounting.

The dissected tissue sections were exposed as a flat sheet between two sliders of inserts. The available tissue area for tilapia and trout was constant and equal to 0.3 cm² (slider no. P2304; Physiologic Instruments). The available tissue area for pig intestinal tissue was 1 cm² (slider no. P2314, Physiologic Instruments). For trout and tilapia, the teleost saline buffer solutions bathing the apical and basal sides of the intestine had a buffer volume of 5 ml and were continuously gassed throughout the experiment with 1% CO₂-99% O₂ for the fish (11). For the pig jejunum sections, the Krebs buffer solution bathing the apical and basal sides of the intestine had a buffer volume of 5 ml and were continuously gassed throughout the experiment with 5% CO₂-95% O₂ (7).

The transepithelial voltage and passing current set across the tissue was measured via agar bridges (3.5% agar in 3 M KCl) and Ag/AgCl reference electrodes. These electrodes were connected to leads that lead to the voltage/current clamp. The pulse was kept constant at 1 mV. A recirculating bath capable of chilling and heating controlled the temperature of the jacketed chambers, with tissues examined at fish housing temperatures. The temperature used was matched to housing temperature for fish and body temperature for pigs; tilapia was 25–26°C, trout was 11-12°C, and pigs was 36–37°C. Needle valves were also present for the adjustment of gas flow into the chambers. This Ussing chamber setup was used to generate the flux and electrogenic data. The specifics of each experiment are explained further below.

Once mounted, the intestinal sections were allowed to reach a steady baseline current for 30–40 min, followed by addition of 1 µCi of 3-O-methyl-d-[¹⁴C]glucose and 1 µCi [³H]inulin to the apical side of the intestinal tissues. An initial 100-µL “hot” sample was taken from the apical (“hot”) side for 3-O-methyl-d-[¹⁴C]glucose and [³H]inulin. An isotope equilibration period was performed after, with a 70-min time period for tilapia and trout and a 60-min time period for pig. Following the isotope equilibration period, a nonlabeled d-glucose gradient from 1 to 8,000 mM was used to assess saturation of transport, with equivalent amounts of nonlabeled d-mannitol additions to the basal side to control for osmotic drift. A gradient of 1 to 8,000 mM was used for all three species to compare trout and tilapia with the pig. The sequential addition of glucose and mannitol was added from a supersaturated stock solution that was dissolved in either teleost saline (for trout and tilapia) or Krebs buffer (for pig). The “ionic strength” of the buffers assisted in glucose and mannitol solubility, as the ions in the buffer created thermodynamically favorable environments to assist in the dissolution of the sugars, especially at the higher concentrations (42). The stock solutions were not dissolved in water as it was not possible to dissolve these high concentrations of sugars in water. Additionally, the sequential additions homogenized with mixing within the chambers from the aeration of the O₂-CO₂ gas mixture. [³H]inulin was used as a control to represent tissue integrity through the duration of the experiment. After every increment of d-glucose, there was a wait period of 10 min after which a 500-µL sample was taken from the basolateral/“cold” side. The radioactive samples were placed into 6-mL liquid scintillation vials that contained 4.3 mL of UltimaGoldcocktail solution (PerkinElmer, Waltham, MA). The radioactive samples were counted in a Beckman scintillation β-counter for ¹⁴C and ³H emission (PerkinElmer).

Electrogenic transport of d-glucose was analyzed from the same generated data. The change in short-circuit current (I_sc) from d-glucose concentrations 0–50 mM was analyzed, resulting in electrogenic plateau. The short-circuit current was measured in microamperes (µA/cm²).

Chemicals.

For pharmacological characterization, 1 × 10⁻⁶, 1 × 10⁻⁵, 1 × 10⁻⁴, 1 × 10⁻³, 1 × 10⁻², 0.1, 0.2, and 0.3 mM of dapagliflozin (inhibitor for SGLT2; AdooQ Bioscience, Irvine, CA); followed by 1 × 10⁻³, 1 × 10⁻², 0.1, 1, and 2 mM of phloridzin dihydrate (inhibitor of SGLT1; AdooQ Bioscience), and followed by 1, 2, 3, 4, and 5 mM of phloretin (inhibitor of GLUT2) (AdooQ Bioscience) were used in sequential order in the Ussing chamber as final concentrations (10, 17, 20, 29). The inhibitors were added to the apical side once d-glucose and d-mannitol concentrations in the chambers reached 8,000 mM. The sequential addition of these inhibitors to the apical side was performed on the same tissue preparation. The sampling included 500-µL samples from the “cold” side for each inhibitor dose at 10-min intervals.

Radioisotopes.

One microcurie of 3-O-methyl-d-[¹⁴C]glucose (nonmetabolizable analog of glucose) and 1 µCi of [³H]inulin obtained from PerkinElmer were used.

RNA extraction and cDNA synthesis using RT-PCR.

Approximately 1-mg samples of tissue were obtained from the dissection of the intestinal tract and stored in RNAlater RNA Stabilization Solution at −80°C. Whole intestinal segments stripped of fat tissue from tilapia and trout were used, and the intestinal segments from pig were stripped of the serosal layer before being used for RNA extraction. RNA was isolated using the TRIzol reagent, and cDNA was synthesized through reverse transcription PCR using the qScript cDNA SuperMix (Quanta Biosciences, Gaithersburg, MD). The reaction was run as: incubation for 5 min at 25°C, then 30 min at 42°C, and the final 5 min at 85°C. The cDNA samples were stored at −80°C for subsequent use in quantitative (q)PCR.

Genomic identification of SLC2A2 (GLUT2) gene.

A detailed BLAST+ (Basic Local Alignment Search Tool) application was used to identify all the annotated GLUT2 transporters in both tilapia, trout, and pig. The human GLUT2 transporter sequence retrieved from the National Center for Biotechnology Information (NCBI) website (https://www.ncbi.nlm.nih.gov/) were used in a “blastn” command to search for similar mRNA sequences in the tilapia, trout, and pig genome (43). The expect value (e-value) was used to assess the significance of the match, where an e-value close to zero was considered significant and an e-value higher than 10^-15 was considered as a nonsignificant match.

Gene transcript expression levels by quantitative polymerase chain reaction.

qPCR was performed on the RT-PCR reaction using the GoTaqPCR Master Mix containing SYBRGreen1 (Promega, Madison WI). The Bio-Rad QPCR system was used to perform these reactions. The total volume of the PCR reaction was 12.5 µL. A total of 40 cycles of qPCR was performed, with each cycle consisting of GoTaq Hot Start Polymerase activation at 95°C for 2 min, and then denaturation at 95°C for 15 s, followed by annealing/extension at the primer’s hybridization temperature (59°C) for 1 min. The dissociation step was at 60–95°C. The amplicon sizes for qPCR products were as follows: 137 bp for tilapia, 98 bp for trout, and 102 bp for pig. Serial dilutions of cDNA were also used to generate a standard curve for each target gene, where the efficiency of each primer was calculated. The efficiencies for the tilapia, trout, and pig primers ranged from 1.9 to 2.1. Tilapia, trout, and pig α-elongation factor 1 (EFα1) was used as the housekeeping gene and was used to normalize the relative mRNA expression levels of GLUT2 for each species. The method of analysis used to represent the relative expression levels was calculated using delta C_T, where the amount of the gene of interest was normalized to the amount of EFα1 in that sample (34).

The primers for pig, trout, and tilapia were designed using the Integrated DNA Technologies website (https://www.idtdna.com/site), where the sequences, and their accession numbers are presented in Table 1. The primers were designed using PrimerQuest (from Integraded DNA Technologies) to span exon-exon junctions and subsequently checked for cross reaction using the NCBI’s Nucleotide-BLAST program. Additionally, the primers were designed to target all the variants of the genes presented here. Primer-dimers were assessed using the melting curve, where one peak was obtained around 79–80°C for all primers used. Finally, each PCR product was sent for sequencing (Macrogen), where the correct sequences were obtained upon cross-checking with the NCBI BLAST program for nucleotide sequences.

Table 1.

Nile tilapia, rainbow trout, and pig primer sequences used for quantitative PCR

Species/Gene	Forward Primer (5′–3′)	Reverse Primer (5′–3′)	Accession No.
Nile Tilapia
GLUT2	CCTCCTTCTTGGTTGGATTT	CAGACGTCAGACCACAATAG	XM_003442884.5
EFα1	CTCTTCTACCGTCGGATTAC	ATTGACTCCCTCGTAGAAAC	XM_005476483.3
Rainbow Trout
GLUT2	CAACCATTGGAGTAGGAGTTC	ACATAGGACATCCAGGAGTAG	XM_021610908.1
EFα1	AGCGAGCTCAAGAAGAAG	GACCAAGAGGAGGGTATTC	NM_001124339.1
Pig
GLUT2	GGACTTGTGCTACTGGATAA	CGTGGTCCTTGACTGAAA	NM_001097417.1
EFα1	GCGAGTGGGATTCTGTTA	GGTGCCTGTCATCTTCTT	XM_013987376.1

EFα1, elongation factor 1; GLUT2, glucose transporter type 2.

Statistical Analyses

The mucosal-to-serosal flux (J_ms) of 3-O-methyl-d-glucose was calculated from the protocol adapted from Schultz and Zalusky (35). The ¹⁴C and ³H emissions from the 500-µL cold samples taken after each concentration increment and inhibitor concentration were recorded as counts per minute (cpm). The J_ms was calculated as follows: Φ_ms^C (J_ms) = ν_s(p_s2 – cp_s1)/(Δtp_m*A), where p_s is counts per minute per cubic centimeter in the serosal reservoir, ν_s is the volume of buffer in the serosal side (in cm³), A is the area of tissue exposed, Δt is the time interval between samples (in hours), p_m* is the specific activity of 3-O-methyl-d-[¹⁴C]glucose in the mucosal solution (in cpm/µmol), and Φ_ms^C is the unidirectional ¹⁴C flux from mucosa to serosa (in µmol·cm⁻²·h⁻¹).

For the mucosal-to-serosal flux of glucose, J_max (represented in µmol·cm⁻²·h⁻¹) and K_m (represented in mM) values were determined for each species’ intestinal sections using GraphPad Prism 8 (GraphPad Software). All three sections followed Michaelis-Menten kinetics (F test with Sigmoidal/Hill fit, P < 0.05 for Michaelis-Menten fit, Fig. 1, A–C), where the equation for Michaelis-Menten is Michaelis-Menten fit: Y = J_max × X/(K_m + X).

Fig. 1.

Kinetics for mucosal-to-serosal flux (J_ms) 3-O-methyl-d-[¹⁴C]glucose flux and [³H]inulin for tilapia proximal intestine (A; fish: n = 15) with J_max and K_m represented for illustration, trout midgut intestinal section (B: fish: n = 17), and porcine jejunum (C; pig: n = 6). The percentage increase of J_ms for the represented concentration comparisons between species (D). I_sc, short-circuit current. Data are represented as means ± SE. *P < 0.05, significance differences for those concentration comparisons within pig (one-way ANOVA).

For the electrogenic transport of d-glucose, the V_max (represented in µA/cm²) and K_m (represented in mM) values were determined for each species’ intestinal segments using GraphPad Prism 8 (GraphPad Software). Tilapia proximal intestine and trout midgut segments followed Michaelis-Menten kinetics (F test, P < 0.05, Fig. 2, A and C), and porcine jejunal segment followed Sigmoidal/Hill kinetics (F test, P < 0.05, Fig. 2E): Michaelis-Menten fit: Y = V_max × X/ (K_m + X) and Sigmoidal/Hill fit: Y =  V_max × X^h/(K_d^h + X^h).

Fig. 2.

Kinetics for electrogenic transport of glucose in tilapia proximal intestine (A; fish: n = 12) with representative trace (B), trout midgut intestinal section (C; fish: n = 12) with representative (D), and porcine jejunum (E; pig: n = 6) with representative trace (F). Data are represented as means ± SE.

A one-way ANOVA was performed between each species to determine significant differences between J_max and K_m values for J_ms (Table 2). All data met parametric assumptions, being normally distributed and exhibited homogeneity of variance. After ANOVA, pairwise comparisons were made using Tukey’s posteriori tests, as appropriate. P < 0.05 was considered significantly different. A one-way ANOVA was used to assess log-transformed (log10) data between each species to determine significant differences between V_max and K_m values for ∆I_sc (Table 2). After ANOVA, pairwise comparisons were made using Tukey’s posteriori tests, as appropriate. P < 0.05 was considered significantly different. The percentage increase of J_ms from the 50-mM increment to the 5- to 8-M increments were calculated for each species, and a one-way ANOVA was performed to determine significance between the compared increments. After ANOVA, pairwise comparisons were made using Tukey’s posteriori tests, as appropriate. P < 0.05 was considered significantly different. For the significance between J_ms and ∆I_sc K_m values for each species, a Student’s t test was performed. P < 0.05 was considered significantly different.

Table 2.

J_ms (J_max and K_m) and ∆ I_sc (V_max and K_m) values for Nile tilapia, rainbow trout, and pig

		Jms		ΔIsc
Species	Intestinal Section	Jms, µmol·cm−2·h−1	Km, mM	Vmax, µA/cm2	Km, mM
Nile tilapia	Proximal intestine	0.009 ± 0.001A	1.6 ± 0.4A	64.6 ± 9.4A	1.8 ± 0.5A
Rainbow trout	Midgut	0.002 ± 0.0003B	3.3 ± 0.7A	2.3 ± 0.7B	2.8 ± 0.9A
Pig	Jejunum	0.004 ± 0.0005B	1,561.4 ± 185.6B*	11.9 ± 2.4C	4.3 ± 0.6B*

Data are presented as means ± SE. J_max and K_m values are represented for Nile tilapia, rainbow trout, and pig mucosal-to-serosal flux (J_ms). Area of intestinal segment was 0.3 cm² for both fish and 1 cm² for pig. V_max and K_m values are represented for Nile tilapia, rainbow trout, and pig electrogenic transport of glucose (∆I_sc). There were no significant differences for J_max values between rainbow trout and pig or for K_m values between Nile tilapia and rainbow trout. A Tukey’s test after one-way ANOVA was used.

^*Significance between J_ms and ΔI_sc K_m values.

^A,BP < 0.05, significant differences for J_max values found between species: Nile tilapia vs. pig and Nile tilapia vs. rainbow trout.

^A,BP < 0.05, significant differences for K_m values found between species: rainbow trout vs. pig and Nile tilapia vs. pig.

^A,BP < 0.001, significant differences for V_max values found between species: Nile tilapia vs. rainbow trout.

^A,CP < 0.001, significant differences for V_max values found between species: Nile tilapia vs. pig.

^B,CP < 0.001, significant differences for V_max values found between species: rainbow trout vs. pig.

To designate whether the transport was super high-affinity (sHa), high-affinity (Ha), low-capacity (Lc), super low-capacity (sLc), and high-capacity (Hc) required statistically comparing K_m and V_max values between the three species. For example, if the K_m values were low and statistically significant from each other, then the lower value would be designated as sHa, and the other value as Ha (Table 2). Similarly for the V_max values, if the values were statistically different from each other, then the lowest value would be designated as sLc (Table 2). The values following this in increasing order would be designated as low-capacity and high-capacity (Table 2). Altogether, this method was used to determine the characterization of total glucose flux and electrogenic glucose transport in all three species.

For the inhibitors, the percent activity remaining was calculated from the division of the previous drug increment (the concentration where the transporter is 100% uninhibited by that specific inhibitor) and the resultant drop in mucosal-to-serosal flux (J_ms) from the addition of each drug concentration. Parametric one-way ANOVAs were performed between the 100% and each drug response to determine significance. After ANOVA, pairwise comparisons were made using Tukey’s posteriori tests, as appropriate. Again, P < 0.05 was considered significant. The K_i values for the inhibitor responses were calculated using the “One-Site Fit LogIC50” from GraphPad Prism 8 (GraphPad Software). This fit follows the equation: Y = Bottom + (Top-Bottom)/[1 + 10^(X-LogIC50)]. Student’s t tests were performed to determine significance between K_i values for that inhibitor between species.

For the qPCR analyses, relative expressions were determined by the Bio-Rad system, where the threshold cycle was established using the housekeeping gene EFα1. All statistical analyses were performed on SigmaPlot (Systat Software, San Jose, CA).

RESULTS

Kinetic Characterization of 3-O-Methyl-d-[¹⁴C]Glucose in Nile Tilapia, Rainbow Trout, and Pig

In tilapia, trout, and pig, the mucosal-to-serosal flux (J_ms) of 3-O-methyl-d-[¹⁴C]glucose was measured in Ussing chambers, and followed Michaelis-Menten kinetics in all three species. Tilapia demonstrated a significantly higher capacity (higher J_max) of glucose flux compared with trout and pig (Table 2 and Fig. 1; P < 0.001). There were no significant differences in J_max values between trout and pig, indicating similar capacities in overall whole tissue glucose absorption. The K_m value in pig was significantly higher (over 400-fold) compared with tilapia and trout, with no significant differences between tilapia and trout K_m values (Table 2 and Fig. 1; P < 0.001). The higher K_m value in pig suggests an overall dominating low-affinity glucose transport system. Altogether, the kinetic values present a super high-affinity, high-capacity (sHa/Hc) total glucose transport system in tilapia, a super high-affinity, low-capacity (sHa/Lc) total glucose transport system in trout, and a low-affinity, low-capacity (La/Lc) total glucose transport system in the pig (Table 2).

Finally, the percent increase of J_ms in response to glucose increments was assessed in each species (Fig. 1D). No significant differences in J_ms increase occurred for tilapia and trout, but significant increases from 50 mM to 5,000, 6,000, 7,000, and 8,000 mM were observed for pig, demonstrating continued glucose absorption above 50 mM (above 300%; Fig. 1D; P < 0.001).

Kinetic Characterization of Electrogenic Transport of d-Glucose in Nile Tilapia, Rainbow Trout, and Pig

In tilapia, trout, and pig, the electrogenic transport of d-glucose was measured in Ussing chambers, and followed Michaelis-Menten kinetics in tilapia and trout, and Sigmoidal/Hill kinetics in the pig. The electrogenic transport of glucose was measured as a change in short-circuit current, and represented by ∆I_sc. Similarly to the flux, tilapia demonstrated a significantly higher capacity (Hc; higher V_max) of electrogenic glucose transport (∆I_sc) compared with trout and pig (Table 2 and Fig. 2; P < 0.001). Additionally, there were further differences in V_max values between trout and pig, revealing a super low-capacity (sLc) in trout and a low-capacity (Lc) electrogenic glucose transport system in the pig (Table 2 and Fig. 2; P < 0.001). The K_m ∆I_sc values were significantly different between tilapia and pig, revealing a super high-affinity (sHa) system in tilapia and a high-affinity (Ha) system in the pig (Table 2 and Fig. 2; P < 0.05). Similarly, the K_m ∆I_sc values were significantly different between trout and pig, revealing a super high-affinity (sHa) system in trout and a Ha system in the pig (Table 2 and Fig. 2; P < 0.05). Altogether, the electrogenic transport (∆I_sc) of glucose is demonstrated by a sHa/Hc system in tilapia, a sHa/sLc system in trout, and a Ha/Lc system in pig.

Finally, the significance between J_ms and ∆I_sc K_m values for each species was assessed to reveal only a significant difference in the pig (Table 2, P < 0.001). Here, the J_ms K_m value was ~363-fold higher than the K_m value from ∆I_sc in the pig (Table 2). More importantly, the porcine jejunum has a different ∆I_sc K_m value from the J_ms K_m value, where this difference is not present in both tilapia and trout (Table 2).

Inhibition of Whole Intestinal Tissue Glucose Flux

A series of inhibitors for sodium-dependent glucose transporters (SGLTs) and glucose transporter type 2 (GLUT2) were used in succession on the apical side to inhibit total glucose flux across the intestinal segments of tilapia, trout, and pig. The sequential addition of each inhibitor, along with its dose response, was in the order of dapagliflozin (SGLT2 inhibitor), phloridzin dihydrate (SGLT1 inhibitor), and phloretin (GLUT2 inhibitor), which was added to the same tissue in each species. This sequential addition was performed after glucose flux reached 8,000 mM in all three species. Figures 3, 4, and 5 illustrate the percent activity remaining after successive addition of each inhibitor in each species.

Fig. 3.

Percent activity remaining of mucosal-to-serosal flux (J_ms) in Nile tilapia by dapagliflozin (A; 0.000001, 0.00001, 0.0001, 0.001, 0.01, 0.1, 0.2, and 0.3 mM; fish: n = 11), followed by addition of phloridzin dihydrate (0.001, 0.01, 0.1, 1, and 2 mM; fish: n = 14), and followed by phloretin (C; 1, 2, 3, 4, and 5 mM; fish: n = 12). *P < 0.05, significance from 100% (where 100% represents transport is 100% uninhibited by that inhibitor), which is presented as the dotted line (one-way ANOVA).

Fig. 4.

Percent activity remaining of mucosal-to-serosal flux (J_ms) in rainbow trout by dapagliflozin (A; 0.000001, 0.00001, 0.0001, 0.001, 0.01, 0.1, 0.2, and 0.3 mM; fish: n = 12), followed by addition of phloridzin dihydrate (B; 0.001, 0.01, 0.1, 1, and 2 mM; fish: n = 9), and followed by phloretin (C; 1, 2, 3, 4, and 5 mM; fish: n = 15). The 100% represents transporter is 100% uninhibited by that inhibitor, which is presented as the dotted line.

Fig. 5.

Percent activity remaining of mucosal-to-serosal flux (J_ms) in pig by dapagliflozin (A; 0.000001, 0.00001, 0.0001, 0.001, 0.01, 0.1, 0.2, and 0.3 mM; pig: n = 6), followed by addition of phloridzin dihydrate (B; 0.001, 0.01, 0.1, 1, and 2 mM; pig: n = 5), and followed by phloretin (C; 1, 2, 3, 4, and 5 mM; pig: n = 4). *P < 0.05, significance from 100% (where 100% represents transport is 100% un-inhibited by that inhibitor), which is presented as the dotted line (one-way ANOVA).

Nile tilapia flux inhibition.

Tilapia had significant inhibition with dapagliflozin from 0.001 to 0.3 mM, with ~60% activity remaining by 0.3 mM dapagliflozin (Fig. 3A; P < 0.05). Similarly, significant inhibition was observed with phloridzin dihydrate, with ~70% activity remaining by the 2 mM concentration (Fig. 3B, P < 0.05). After phloridzin dihydrate, phloretin did not reveal significant inhibition with any of its concentrations (Fig. 3C). Finally, K_i values were calculated for these inhibitors (Table 3). Inhibition by both dapagliflozin and phloridzin dihydrate strongly support involvement of SGLT transporters.

Table 3.

K_i values for Nile tilapia, rainbow trout, and pig

Species	Dapagliflozin Ki, Mm	Phloridzin Dihydrate Ki, mM	Phloretin Ki, mM
Nile tilapia	0.007 ± 0.002A	1.2 ± 0.7A	ND
Rainbow trout	ND	ND	ND
Pig	0.004 ± 0.001A	0.5 ± 0.3A	3.6 ± 1.6

Data are presented as means ± SE. K_i values are represented for Nile tilapia, rainbow trout, and pig. ND, nondetectable data.

^ASame letter superscripts indicate no significant differences between Nile tilapia and pig for dapagliflozin and phloridzin dihydrate (Student’s t test, P < 0.05).

Rainbow trout flux inhibition.

No significant inhibition was observed for dapagliflozin, phloridzin dihydrate, and phloretin (Fig. 4, A–C). Therefore, no K_i values were calculated for any of the inhibitors (Table 3).

Pig flux inhibition.

Dapagliflozin revealed significant inhibition from 0.0001- to 0.3-mM concentrations, with ~50% activity remaining at 0.3 mM (Fig. 5A, P < 0.05). Similarly, phloridzin dihydrate revealed significant inhibition from 0.01 mM to 2 mM concentrations, demonstrating ~60% activity remaining at 2 mM phloridzin dihydrate (Fig. 5B, P < 0.05). In contrast to tilapia, pig revealed significant inhibition with phloretin, from 3 to 5 mM (Fig. 5C, P < 0.05). Additionally, ~70% activity remained with phloretin at 5 mM (Fig. 5C). Finally, K_i values were calculated for these inhibitors (Table 3). No significant differences in K_i values were found with dapagliflozin and phloridzin dihydrate between tilapia and pig, indicating strong contribution of SGLTs in both species (Table 3). Additionally, the strong inhibition by phloretin suggest involvement of GLUT2 transporters.

Gene Expression

In a CLUSTALW analysis, the nucleotide identity of GLUT2 was compared between the three species and relative to tilapia GLUT2 (SLC2A2) were 70% GLUT2 (SLC2A2) in trout and 62% GLUT2 (SLC2A2) in pig. A phylogenetic tree was generated between GLUT2 nucleotide sequences of tilapia, trout, and pig (Fig. 6A).

Fig. 6.

A: a phylogenetic tree of glucose transporter type 2 (GLUT2) in tilapia, trout, and pig. The alignment program CLUSTAL W and the phylogenetic display program MEGA7 were used to generate the tree (A). Relative expression levels of GLUT2 transporters in Nile tilapia proximal intestine (B; n = 10), rainbow trout midgut intestinal section (C; n = 10), and pig jejunum (D; n = 10) using quantitative PCR. Expression levels were normalized against α-elongation factor 1 (EFα1) housekeeping gene in each species.

The GLUT2 (SLC2A2) transporter was identified for tilapia, trout and pig genome using the BLAST+ application (Table 1). The expression levels of GLUT2 in whole intestinal segments were compared among the different species using quantitative real-time PCR (Fig. 6, B–D).

DISCUSSION

For the first time, we have characterized the total mucosal-to-serosal flux of glucose transport in tilapia, trout, and pig and compared it against the electrogenic transport of glucose. Differences in affinities (K_m “fingerprints”) revealed major differences in the pathways involved in glucose absorption between the terrestrial and teleost species studied here. Along with inhibitor and gene expression analyses, we suggest apical GLUT2 transporter involvement in the pig but not in tilapia and trout. It would appear apical GLUT2 transporter involvement in the pig likely contributes to continued glucose absorption above the capacity of SGLTs. Altogether, the pig demonstrates a two-absorptive pathway for total glucose transport, whereas both fish species demonstrate a one-absorptive dominating pathway for total glucose transport.

Total Glucose Flux and Electrogenic Glucose Absorption Reveals Species Differences in Pathways Involved in Glucose Absorption

Kinetic characterization of total intestinal glucose absorption (J_ms) between tilapia, trout, and pig revealed an overall sHa/Hc transport system in tilapia, a sHa/Lc transport system in trout, and a La/Lc transport system in pig (Fig. 1, A–C and Table 2). Comparatively, electrogenic transport (∆I_sc) of glucose revealed a sHa/Hc system in tilapia, a sHa/sLc system in trout, and a Ha/Lc system in the pig (Table 2). The kinetic characterizations in tilapia and trout represent similar kinetic systems in our previous work (37). Altogether, the K_m fingerprints from both J_ms and ∆I_sc strongly support a one-absorptive pathway in fish and a two-absorptive pathway in pig.

J_ms and ∆I_sc K_m Fingerprints Define One-Absorptive Pathway in Tilapia and Trout

Specifically, no significant differences were found with the K_m values between mucosal-to-serosal flux (J_ms) and the electrogenic transport (∆I_sc) of glucose in both tilapia and trout (Table 2). This suggests similar transporters responsible for total glucose flux (J_ms) and electrogenic glucose transport (∆I_sc) in both tilapia and trout. The differences in capacity of total glucose flux (J_ms) and electrogenic transport (∆I_sc) of glucose reveals the higher capacity for glucose absorption in tilapia than trout. This is in line with previously described differences in the expression of SGLTs, where SGLT expression is higher in tilapia than trout (37). Additionally, the low K_m values for tilapia and trout in the millimolar range for glucose flux (J_ms) and electrogenic glucose (∆I_sc) transport indicate that transporters involved in glucose absorption are generally high-affinity, suggesting the lack of GLUT2 contribution (15, 17, 27). The lack of percentage increase in J_ms after 50 mM glucose for both fish species further suggests only the involvement of SGLTs and not GLUT2 (Fig. 1D). Finally, the similarity in K_m values for both J_ms and ∆I_sc following previously described expression of SGLTs strongly suggests they are the major route of glucose absorption in these two species, indicating a one-absorptive pathway (37).

This is not surprising as both fish species have evolved to use lower dietary glucose than mammals, as previously described (14, 30). This is particularly true for carnivorous fish like rainbow trout, where glucose use is not the primary energy source compared with omnivorous fish and mammals (22, 31). This is reflected by the lower V_max in total glucose absorption (J_ms), as well as electrogenic transport (∆I_sc) of glucose in trout compared with tilapia (Fig. 1, A and B, and Table 2). It has also been demonstrated in the omnivorous fish like the black bullhead catfish that glucose usage is lower than mammals, which may also be true for the omnivorous tilapia (18, 31). Therefore, the possibility of apical GLUT2 absorption for tilapia and trout may not exist or is not needed. The very low K_m values in comparison to the pig support this notion.

J_ms and ∆I_sc K_m Fingerprints Define Two-Absorptive Pathway in Pig

In contrast, the pig total glucose transport (J_ms) is a La/Lc system, with a K_m in the molar range that is over 400-fold higher than both tilapia and trout (Table 2). Comparatively, the electrogenic transport (∆I_sc) of glucose is a Ha/Lc system with a K_m value from ∆I_sc that is 363-fold lower than the K_m value resulting from J_ms (Table 2; compare 4.3 with 1,561 mM). The difference in affinities, or the K_m “fingerprint,” between total glucose absorption (J_ms) and electrogenic glucose (∆I_sc) transport in the pig indicates a different absorptive pathway of glucose absorption compared with tilapia and trout. More specifically, this very low-affinity for total glucose transport (J_ms) in the pig has been described in other mammals, such as mice and rats (17). This system was attributed to apical GLUT2 insertion that allowed for higher glucose absorption, thus increasing the overall K_m value of total glucose transport (16, 17, 41). However, the K_m values for GLUT2 reported in these studies ranged from 17 to 56 mM and were dependent on the glucose range and techniques used (8, 17). These K_m values are over 20-fold lower than the K_m value presented in this study, possibly due to differences in techniques such as glucose range, type of assay, and species used (rodents versus pig) (17). However, studies specifically looking at GLUT2 transport found it did not saturate even after 100 mM, suggesting a very large K_m (15–17, 44). Thus the actual K_m of GLUT2 and species differences has not been fully elucidated. Finally, the total glucose transport (J_ms) in the pig, which continued after 50 mM glucose, in comparison with tilapia and trout, exhibited a significant percentage increase of J_ms at the higher concentrations (Fig. 1D). Therefore, the presence of apical GLUT2 transporters may explain the higher saturation of total glucose transport present in the pig, shifting the K_m value away from SGLTs. Additionally, the presence of GLUT2 was supported by pharmacological inhibition.

Inhibition of Glucose Flux Defines Different Transport Pathways between Species

Inhibitor responses from dapagliflozin, phloridzin dihydrate, and phloretin demonstrated differences between the fish species here and the pig (Figs. 3, 4, and 5). Dapagliflozin and phloridzin dihydrate in tilapia demonstrated the involvement of SGLTs for total tissue glucose absorption (J_ms) (Figs. 3, A and B). This has been similarly demonstrated in a previous study that focused on sodium-dependent glucose transport in tilapia and found inhibition with both dapagliflozin and phloridzin dihydrate (37). Since most of inhibition was achieved by SGLT inhibitors, it suggests that majority of transport in tilapia is through SGLTs. Additionally, the lack of increase in total glucose absorption (J_ms) after 50 mM, and similar affinities between electrogenic glucose transport (∆I_sc) and total glucose flux (J_ms), further support the transport of glucose in tilapia may be predominantly through SGLTs/SLC5A transporters. Finally, the lack of phloretin response in tilapia suggests a lack of apical GLUT2 involvement with total glucose absorption (J_ms) (Fig. 3C). Additionally, lack of inhibition with phloretin can also be due to stereospecificity of the inhibitor. However, the differences in GLUT2 sequence between tilapia and pig (Fig. 6A) can influence the lack of inhibition by phloretin on tilapia GLUT2 transporters. In contrast, trout did not exhibit any significant responses to any of the inhibitors (Fig. 4, A–C). It could be argued that the lack of significant inhibition to total tissue glucose absorption (J_ms) in trout may be due to the low temperature used to conduct the flux experiments (~11–12°C, ideal housing temperature for poikilothermic trout), some of which have been previously described (37). Phloretin has been used successfully in other fish species, specifically to inhibit glucose transport in the liver which contains GLUT2 transporters (20a, 26, 39a). Its lack of response in both tilapia and trout here is likely due to the lack of GLUT2 transporters present on the apical side, since the inhibitors were administered to the apical side of the intestinal segments.

Similar to tilapia, the pig responded to inhibition by dapagliflozin and phloridzin dihydrate, indicating the contribution of SGLTs in total tissue glucose absorption (J_ms) (Fig. 5, A and B). However, in contrast to tilapia and trout, significant inhibition with phloretin was observed (Fig. 5C). GLUT2 phloretin sensitivity has been shown in rat jejunum and Xenopus laevis oocyte expression systems (13, 16, 40, 44). This supports GLUT2 involvement in the pig, giving insight into the large shifted K_m in comparison to the fish species here. Additionally, SGLT1 is not sensitive to phloretin, further supporting the fact that GLUT2 transporters may be present in pig jejunum (17, 21).

Gene Expression Support Functional Apical GLUT2-Transporter Involvement in Pig

Gene expression analysis of GLUT2 in whole intestinal segments from all three species revealed GLUT2 expression in all three species (Fig. 6, B–D). The high expression in the pig adds to the kinetic and inhibitor analyses, suggesting a large pool of GLUT2 transporters (normalized to the housekeeping gene EFα1) contributing to the overall process of total glucose absorption (J_ms). While we cannot confirm that GLUT2 was apically expressed in this study, this plausible GLUT2 contribution to total glucose absorption (J_ms) may indicate that the pig also has a GLUT2 apical translocation mechanism, which has been shown previously to assist with high glucose absorption in mice and rats (16, 17). In addition to GLUT2 transporters, the pig jejunum also involves SGLT transporters, based on the responses to the inhibitors and similar K_i values to tilapia inhibition. Therefore, the possible involvement of SGLTs and GLUT2 in pig total glucose absorption (J_ms), as well as the drastic differences in K_m “fingerprints” between total (J_ms) and electrogenic transport (∆I_sc) of glucose, strongly support a two-absorptive pathway for total glucose absorption (J_ms).

Conclusion

Total glucose absorptive flux (J_ms) presents a sHa/Hc transport system in tilapia, a sHa/Lc transport system in trout, and a La/Lc transport system in pig. Similarly, electrogenic glucose (∆I_sc) transport revealed a sHa/Hc system in tilapia, a sHa/sLc system in trout, and a Ha/Lc system in the pig. Overall, glucose absorption in tilapia and trout most likely occurs through SGLTs, with a lack of apical GLUT2 involvement, indicating a dominating one-absorptive pathway for total glucose absorption. In contrast, glucose absorption in the pig jejunum may be attributed by SGLTs, with functional apical GLUT2 involvement, indicating a two-absorptive pathway for total glucose absorption. Given these differences in transporter involvement, this study demonstrates continued glucose absorption at higher concentrations in the porcine jejunum, whereas this is not observed in tilapia and trout. These transport pathways observed in the mammalian pig may be an evolutionary adaptation different from fish, which may not be exposed to a diet high enough in free glucose to allow facilitative apical glucose absorption. Finally, this contrast between species highlights the novel mechanism of total glucose absorption in the porcine model and validates the observations.

Limitations

There are limitations present in this study. One of the limitations is the results obtained from the gene expression analyses represent assumptions, since mRNA and protein expression is not always correlated. However, performing Western blots using fish antibodies for SGLTs and GLUTs is not possible, as fish antibodies are not available; thus comparisons cannot be made. Additionally, GLUT2 expression was conducted using whole intestinal tissues, rather than scraping the apical portion of the intestine. Therefore, possible GLUT2 expression from the basolateral membrane may also contribute to the overall expression (hence seeing some GLUT2 expression in both fish species). However, our conclusions of GLUT2 presence on the apical side of the pig jejunum is based on the support of using inhibitors only on the apical side of the intestine and observing significant inhibition. Another assumption made in this study is that if GLUT2 was present on the apical side of fish, it would transport similarly to mammalian GLUT2. The fourth limitation presented in this study would be the cross-species comparison of GLUT2 expression between tilapia, trout, and pig. With the differences in GLUT2 gene sequences between the species, it resulted in using different primers which generated different amplicons. Therefore, statistically comparing GLUT2 expression between the species would be complicated. Finally, in a physiological environment, glucose and mannitol concentrations would never reach as high as 8 M, as presented in this study. However, for the experiments presented here, to ensure maximum saturation was reached, a glucose gradient had to be run up to 8 M to achieve an accurate fit with appropriate K_m and V_max values, which are essential for kinetic comparative analyses. The glucose flux in the pig saturates around 5–6 M glucose in vitro, which demonstrates that the pig has the capability of absorbing very high glucose concentrations, above the capacities of SGLTs. However, the assumption is not made that the pig normally absorbs concentrations of 5–6 M glucose along its jejunum in vivo. Overall, the biological meaning set out in this study was to demonstrate that the pig has the capacity to absorb glucose concentrations much higher than the capacities of SGLTs, suggesting a two-absorptive pathway for glucose absorption. Finally, both fish species were run to a gradient of 8 M to compare with the pig results and serve as a control for each.

Perspectives and Significance

Here, we present the first report comparing the kinetic characterization of total tissue glucose absorption (J_ms) and electrogenic glucose (∆I_sc) absorption between tilapia proximal intestine, trout midgut, and pig jejunum intestinal segments. The K_m fingerprint between the mucosal-to-serosal flux (J_ms) and electrogenic transport (∆I_sc) of glucose in the pig is over 300-fold different. In contrast, this is not seen in both tilapia and trout. Therefore, with supporting data, the pig most likely demonstrates a two-absorptive pathway for glucose, whereas only a one-absorptive pathway dominates in both tilapia and trout. The two-absorptive pathway in the pig allows it to have continued glucose absorption at high concentrations. This suggests that mammals may have evolved to absorb higher amounts of glucose than aquatic species. Altogether, elucidating this mechanism of high glucose absorption in the mammalian model pig can allow one to consider other routes to control glucose uptake in diseases such as diabetes. More specifically, when attempting to control glucose uptake from the intestine by pharmacological inhibition, more than one transport pathway needs to be considered.

GRANTS

This research was supported by Natural Sciences and Engineering Research Council of Canada Discovery Grant 371364-2010 (to M. E. Loewen), Natural Sciences and Engineering Research Council of Canada Collaborative Research and Development Grant CRDPJ-446912-13 and the Saskatchewan Pulse Crop Development Board (to M. E. Loewen and L. P. Weber).

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

M.E.L. conceived and designed research; M.E.L. and M.S. performed experiments; M.S., C.E., and K.L. analyzed data; M.E.L. and M.S. interpreted results of experiments; M.S. prepared figures; M.E.L. and M.S. drafted manuscript; M.S., L.P.W., and M.E.L. edited and revised manuscript; M.S., L.P.W. and M.E.L. approved final version of manuscript.