Ancestral genomic duplication of the insulin gene in tilapia: An analysis of possible implications for clinical islet xenotransplantation using donor islets from transgenic tilapia expressing a humanized insulin gene

ABSTRACT

Tilapia, a teleost fish, have multiple large anatomically discrete islets which are easy to harvest, and when transplanted into diabetic murine recipients, provide normoglycemia and mammalian-like glucose tolerance profiles. Tilapia insulin differs structurally from human insulin which could preclude their use as islet donors for xenotransplantation. Therefore, we produced transgenic tilapia with islets expressing a humanized insulin gene. It is now known that fish genomes may possess an ancestral duplication and so tilapia may have a second insulin gene. Therefore, we cloned, sequenced, and characterized the tilapia insulin 2 transcript and found that its expression is negligible in islets, is not islet-specific, and would not likely need to be silenced in our transgenic fish.

Introduction

Most teleost (i.e., boney) fish have anatomical separation of their endocrine and exocrine pancreata. These fish have one or more large islets called Brockmann bodies (BBs), named after the German ichthyologist Heinrich Brockmann who described these structures in his doctoral thesis De Pancreate Piscium in 1846. BBs played a significant role in the discovery of insulin and in islet-based research up until the mid 1960s when it became possible to isolate mammalian islets.^1,2 Since 1991, our laboratory has been using BBs from the Nile tilapia (Oreochromis niloticus) for islet xenotransplantation studies,³ and we have recently reviewed > 50 of our related publications elsewhere.⁴ In brief, tilapia BBs, which are macroscopically visible, can be inexpensively harvested at high purity without costly and fickle islet isolation procedures.⁵ When transplanted into diabetic athymic nude mice, they provide long-term normoglucemia and mammalian-like glucose tolerance profiles.^3,6,7 Rejection of tilapia BBs in euthymic murine recipients is CD4 T-cell dependent and mechanistically similar to that of porcine islet xenografts.⁸ However, unlike porcine and most other mammalian islets, tilapia are too phylogenetically primitive for their islets to express the dominant xenoantigen, α(1,3)Gal.⁹ Because of the low metabolic needs of tilapia (n.b., tilapia are evolved to thrive in warm stagnant water nearly devoid of dissolved oxygen), tilapia BBs are an order of magnitude more hypoxia resistant than mammalian islets¹⁰ making them ideal for survival and long-term function within encapsulation devices¹¹ which can be further prolonged when combined with co-stimulatory blockade.^12,13

Initially, our intent was to use tilapia BBs simply as an inexpensive source of xenogeneic islets to study the mechanism of xenograft rejection and to gain insights into its prevention. However, because of many of the features listed above, we wondered if it might be possible to use them clinically. However, tilapia BBs secrete tilapia insulin which differs from human insulin by 17 amino acids.¹⁴ Therefore, we made transgenic tilapia bearing BBs that secrete physiological levels of humanized insulin and have now bred these to homozygosity and demonstrated high levels of circulating human insulin for the maximal achievable lifespan of the species (∼9 years).^15,16,17 While these transgenic tilapia, have BBs which simultaneously secrete human and tilapia insulin, we believe that the wild-type tilapia insulin gene, like other fish genes including those of tilapia, can be silenced using CRISPR technology if required.^18,19 In anticipation of clinical islet xenotransplantation studies, we have extensively characterized tilapia BBs (cell biology, morphology, embryonic development, post embryonic growth proliferation, peptides, etc) and their regulation of glucose homeostasis; for a review see⁴. We have also extensively characterized the tilapia insulin gene (NTins1) as described elsewhere.^20,21 What we did not anticipate when we began this work was that tilapia might have 2 non-allelic insulin genes, and if functional, a second tilapia insulin gene product.

It is now generally accepted that the entire genome of ray-fined fish underwent a round of whole duplication during the early stages of evolution.^22-24 This event resulted in the presence of many paralogous genes in the genome of teleost species. One of the retained functional duplicates is the insulin gene that is found in 2 non-allelic copies in the genome of several fish including fugu (Fugu rubripes) and zebrafish (Danio rerio).²⁵

The physiological role of the second teleost insulin remains mostly undetermined. Insulin 2 gene expression has been detected in the zebrafish and rainbow trout embryoes, suggesting a potential functional significance during early development.^26,27 With the exception of a couple of studies in rainbow trout, no data are available concerning insulin 2 gene expression in adult fish.^27,28 In this study, we cloned, sequenced, and characterized the insulin 2 transcript isolated from adult Nile tilapia and analyzed these data in the context of possible future clinical islet xenotransplantation using transgenic tilapia expressing a humanized insulin gene.

Materials and methods

Fish maintenance and tissue isolation

Tilapia maintenance and tissue extraction were performed as previously described.²⁹

Cloning of the partial sequence of the NTins2 gene

Based on the alignment of the available sequences for fish insulin 2 genes, we designed degenerative primers: In2-S1, In2-AS1, In2-S2, In2-AS2, and used them together with the tilapia genomic DNA in 2 consecutive rounds of PCR. The first round was performed with In2-S1and In2-AS1 primers, followed by the second round with In2-S2 and In2-AS2 primers. Sequences of all primers are listed in Table 1. Both PCR mixtures were composed of 75 mM Tris-HCl (pH 8.8), 20 mM (NH₄)₂SO₄, 0.01% Tween 20, 2.5 mM MgCl₂, 200 μM each of dNTPs, 200 nM of each of the primers and 1 U of the Taq DNA polymerase (recombinant) (Fermentas). PCRs were performed in a UNOII Thermocycler (BiometraLtd.) using the following conditions: 50 cycles of denaturing (94°C) for 20 s, annealing (56°C) for 30 s, and elongation (72°C) for 1 min. PCR product was resolved on a 1% agarose gel, extracted and purified using QIAquick gel extraction kit (Qiagen), cloned into the pCRII (Invitrogen), vector and commercially sequenced.

Table 1.

PCR and qPCR primers.

cDNA product	Primer	Primer sequence
β-actin	actFOR	5′-AAGATGAAATCGCCGCACTGGTTG-3′
	actREV	5′-AGGTGTGATGCCAGATCTTCTCCA-3′
NTins2	In2-S1	5′TCCMKCCCAGCAYCTGTG3′
	In2-AS1	5′YRCAGTATCYCTSCAGGTGGT3′
	In2-S2	5′GGYTCMMRCCTGGTGGA3′
	In2-AS2	5′GGCYTRTGGCAGCACTGCTCCAC3′
	In2-S3	5′GAGCCGGACCCACAAGCG3′
	In2-S4	5′CTGGATGTCTGCCTTCTCAACTATATCCC3′
	In2-AS4	5′AAGACATAGCCAGCAAGTATCAGGCC′3
	In2-AS5	5′GTCGCGGCCCGACAAAGCCCTCCA3′
	In2-AS6	5′TCCACAGTCGCTGGTCCTGTCTGGCCC′3
NTins1	In49	5′GAGATGTGGACCCTCTGCTTG3′
	In6	5′TGTAGAAGAAGCCTCTCTCCCC3′

The obtained 898 bp clone contained partial sequence of the NTins2 gene that included: 100 bp of the second exon, 146 bp of the third exon, and 682 bp of the second intron. The exon sequences of this product were used to design NTins2 gene specific primers In2-S3, In2-AS3 and to perform preliminary tissue expression analysis. This experiment revealed the presence of NTins2 mRNA in the total RNA extracted from the pancreatic islets. Therefore, pancreatic islet RNA was used to obtain 5′ and 3′ ends of the NTins2 cDNA employing RACE techniques.

RNA isolation and reverse transcription (RT)

Total RNA was isolated from different tissues using TRIzol reagent (Invitrogen) according to the recommended protocol. To remove contaminating DNA, aliquots of isolated total RNA were treated with 1 U of DNAseI (Invitrogen) and then were reverse transcribed using a SuperScript III reverse transcriptase kit (Invitrogen) and oligo(dT) primer (Invitrogen).

Rapid amplification of cDNA ends (RACE)

To obtain 5′ and 3′ ends of the NTins2 mRNA, 3′ and 5′ RACE were performed using pancreatic islets total RNA and a GeneRacer Kit (Invitrogen) according to the provided protocol. For the 5′RACE, In2-AS5 and In2-AS6 NTins2-specific primers were used in the first and in the second rounds of PCRs respectively. For the 3′RACE, In2-S2 and In2-S3 primers were used in the first and in the second rounds of PCRs respectively. The 527 bp 5′ RACE product and 341 bp 3′RACE product were resolved on a 1% agarose gel, extracted and purified using a QIAquick gel extraction kit, cloned into the pCRII vector and commercially sequenced.

Cloning of full length pancreatic islet NTins2 transcript

To confirm the entire sequence of the NTins2 transcript we performed PCR using pancreatic islet cDNA and In2-S4/In2-AS4 primers that correspond to the 5′ and 3′ terminal sequences of the NTins2 RACE products.

PCR conditions were the same as for the amplification of the NTins2 gene product, except primer annealing was performed at 58°C.

Polymerase chain reaction (PCR) to study tissue-specific expression of the NTins2 genes

Two microliters of the cDNA samples were added in each of 2 PCRs- with primers specific (1) to the tilapia insulin 2 (In2-S4/In2-AS10) and (2) to the tilapia β-actin transcript (actFOR/actREV). Sequences of all primers are listed in Table 1. All PCR were performed in the same conditions as described above, except the elongation was 30 s. PCR products were resolved on 6% polyacrylamide gels, visualized by staining with ethidium bromide and photographed.

Quantitative PCR (qPCR) to analyze the levels of insulin expression in different tissues

Quantitative amplifications of NTins1, NTins2, and β-actin products and data analysis were performed as described by us earlier.²⁹ qPCR primers were the same as for the corresponding PCR (see Table 1).

Results

Sequence analysis of the tilapia insulin 2

The complete NTins2 transcript is 750 bp long and contains 333 bp of the main open reading frame, as well as 295 bp of the 5′ and 122 bp of the 3′ untranslated regions. Interestingly, an additional 12 bp reading frame was detected in the 5′ end of the NTins2 cDNA (113–124 bp). The deduced amino acid sequence of the main reading frame encoded for a protein of 111 aa with MW of 12.6 kDa.

Analysis of the primary protein organization revealed that precursor molecule consists of the 4 typical insulin regions, including a signal peptide (23 aa), B chain (32 aa), C chain (31 aa), and A chain (21 aa). Signal peptide cleavage site was predicted using SignalP software (http://www.cbs.dtu.dk/services/SignalP/). The prohormone convertase processing sites were determined by identification of the consensus sequences recognized by these enzymes (RR or KR)^30,31 and based on the sequence similarity with insulin precursors from other species. The location of the prohormone processing sites and the size of signal peptide, B-, C-, A-chains of the TNins2 were similar to those found in the other known insulin precursors (Fig. 1).

Figure 1.

Alignments of the tilapia, zebrafish and fugu fish insulin precursor sequences. Insulin A and B chains are in bold. Residues important for the peptide folding are underlined with single line and residues essential for the receptor binding are underlined with double line. Prohormone convertases and carboxypeptidase H processing sites are space separated. The second potential prohormone processing site of the fugu fish insulin 2 precursor is shown in italic.

Tissue distribution analysis of the NTins2 cDNA

To examine physiological significance of NTins2 we analyzed expression of the tilapia NTins2 gene in different tilapia tissues including gill, liver, BBs, heart, small intestine, adipose tissue, brain, pituitary gland, spleen, red muscles, and white muscles using end-point RT-PCR (Fig. 2). Insulin-2 transcripts were detected in BBs, brain, red and white muscles, and pituitary gland.

Figure 2.

Tissue-specific expression of the NTins2 gene. Total RNA was isolated from several tilapia tissues: gill (G), liver (L), Brockmann body (BB), heart (H), small intestine (SI), adipose tissue (A), brain (B), pituitary gland (PG) spleen (S) red muscles (RM) and white muscles (WM) and was used in RT-PCR with NTins2-specific and β-actin-specific primers; (−) negative control, (+) positive control.

To confirm our finding and to compare levels of NTins2 mRNA in BB and in non-islet tissues we performed RT-qPCR using total RNA isolated from the same set of tissues of adult tilapia fasted for 72 h (Fig. 3, white bars). We also investigated if glucose (1g/kg injected IP) and/or feeding with protein diet (50% protein/ 20% fat/ 2% fiber diet) could influence expression of NTins2 (Fig. 3, black and gray bars). Fish were fasted for 72 h and then either feed or injected. Tissues were extracted 4 h later.

Figure 3.

Levels of NTins2 expression in tilapia tissues. Effects of protein based food and intraperitoneally injected glucose on the levels of NTins2 expression were examined in the same tissues as described in Fig. 2 using RT-qPCR. Total RNA was isolated from tilapia tissues from: (white bars) fish fasted for 72 h; (black bars) fish fasted for 72 h and then fed 4 h before tissue harvesting; and (gray bars) fish fasted for 72 h and then injected intraperitoneally with 1g/kg glucose 4 h before tissue harvesting. Quantifiable levels of NTins2 mRNA were detected in Brockmann body (BB), brain (B), pituitary gland (PG), and red muscles (RM). Results of the RT-qPCR are presented as the NTins2/actin mRNA ratios and are shown relatively to the NTins2/actin mRNA ratio of the fasted BB sample which is arbitrary set as 1. Data are expressed as mean SE of at least 3 independent experiments.

Similarly to the results of end-point RT-PCR quantitative levels of Insulin 2 mRNA were detected in BBs, brain, pituitary gland and in red muscle. In other tissues levels of NTins2 mRNA were either non detectable or below quantification limits.

In fasted fish the highest level of NTins2 mRNA was detected in the tilapia BB. Lower levels of NTins2 transcript were found in red muscle (28% of the BB mRNA level), followed by pituitary gland (22% of the BB mRNA level) and brain (1% of the BB mRNA level. As shown in Fig. 3, neither food nor glucose had any significant effect on NTins2 mRNA levels in all tissues tested.

Previously we have demonstrated that tilapia insulin 1 gene is actively transcribed in tilapia BBs.^20,21 To assess whether both insulin genes contribute to the production of the circulating protein we compared levels of insulin1 and 2 expression in tilapia BBs. We found that the levels of NTins2 transcription were 10⁷ fold lower than the NTins1 mRNA levels (Fig. 4).

Figure 4.

BB level of insulin 2 mRNA is drastically lower than the level of insulin 1 mRNA. Total RNA was isolated from tilapia BBs and used in RT-qPCR. Results are calculated as NTins1/actin or NTins2/actin mRNA ratios and are shown relatively to the NTins2/actin mRNA level which is arbitrary set as 1. Data are expressed as mean SE of at least 3 independent experiments.

Discussion

The NTins2 sequence possesses the conserved amino acid residues important for correct folding, formation of disulfide bonds, and receptor binding,³² suggesting similarity to the zebrafish and fugu insulin 2 precursors.²⁵ Therefore, it appears that NTins2 precursor could be correctly processed to a bioreactive mature insulin. According to the alignment of the tilapia, zebrafish, and fugu fish preproinsulin sequences, the NTins2 precursor shares 52% identity with the NTins1 precursor, 54% and 75% identity with zebrafish and fugu fish preproinsulin 2 respectively. The higher sequence similarity of the NTins2 precursor to the insulin 2 precursors from other fish species supports the hypothesis that the 2 non-allelic fish insulin genes arose before radiation of the teleost fish.²⁵

The greater divergence among NTins2 precursor and other known insulin precursors was found within the signal and connective peptides, whereas mature NT ins2 (A- and B- chains) showed 72 % identity with the NTins1 and zebrafish insulin 2, and 80% identity with the fugu fish insulin 2.

In adult mammals, the expression of the insulin gene is almost exclusively restricted to the pancreatic islet β-cells. Previously we demonstrated that the insulin 1 gene in tilapia is expressed predominantly in BBs, but also in some non-islet tissues including pituitary gland and brain.²¹ In this study we found that the tilapia NTins2 gene is expressed in BBs, brain, red and white muscles, and pituitary gland. This suggests that in contrast to the β-cell-specific expression of the mammalian insulin gene, expression of the both insulin genes in tilapia is not strictly β-cell-specific.

When we compared levels of insulin 1 and 2 expression in BBs we found that, tilapia BBs transcribe both insulins however the level of the insulin 2 mRNA is 10⁷ fold lower than the level of insulin 1 mRNA, suggesting that the insulin pool in tilapia islets likely consists either exclusively or almost exclusively of the insulin 1 gene product. This premise is further supported by the observation that our previous attempt to biochemically extract and purify all tilapia BB peptides did not yield evidence of a tilapia insulin 2 product.¹⁴

In non-islet tissues, the levels of NTins2 insulin mRNA were exceedingly low. Quantitative levels of Insulin 2 mRNA were detected in brain, pituitary, and red muscle. Clearly, the NTins2 gene is transcribed at low levels, or possibly only in few cells, in non-islet tissues and its transcription is only slightly higher in BBs.

We also tested if glucose injected IP at 1g/kg [n.b., this is a very substantial glucose load for tilapia^33,34] or if protein rich food could induce levels of insulin 2 mRNA in tilapia BB and non-islet tissues. We found that neither glucose nor protein-based food affected levels of insulin 2 expression.

In conclusion, our results revealed that in contrast to most mammalian genomes and similar to the genomes of other teleost fish, the tilapia genome contains 2 non-allelic insulin genes that both encode for potentially bioactive proteins. While the recognition that fish species may have 2 insulin genes is relatively recent, it has been known for over 35 y that rodents possess 2 non-allelic insulin genes.³⁵ Taxonomically, duplication of insulin and other genes have been used to study genome evolution in both fish and rodents.^22-24,36 Thus far, these types of studies have been more comprehensive in rodents (family Muridae) – both in rats and 9 different species of mice, representing 4 different murid subfamilies. Both rats and mice have 2 functional insulin genes, preproinsulin 1 (Ins1) and preproinsulin 2 (Ins2). Both are expressed in islets, are functional, and encode for structurally normal proinsulins. Ins2 appears to be the ortholog of the mammalian insulin gene and Ins 1 is a rodent-specific retrogene. Ins1 appears to have originated from a partially processed reverse-transcribed mRNA of Ins2 and it retains only the first of 2 introns in Ins2. It is believed that Ins1 appeared about 20 million years ago (i.e., before the evolutionary mouse-rat split). In mice, these 2 genes reside on different chromosomes, whereas in rats they are on the same chromosome but >100 MB apart.³⁶ To date, duplication of insulin genes has not been identified in mammals other than rodents. In teleost fish, which as a superorder represent about half of all living vertebrates,²³ there is insufficient data to make any generalizations except that there appears to have been a whole genome duplication 300–400 million years ago,^22-24 and so likely most (maybe all) possess an Ins2 gene. Much work needs to be done in this arena.

In adult tilapia, insulin 2 gene expression occurs in BBs, brain, pituitary gland and muscles. However, levels of NTins2 expression are negligibly low in non-islet tissues, and are only slightly higher in BBs; the physiological significance of such expression is doubtful. The dramatically higher level of insulin 1 expression compared to that of insulin 2 expression in tilapia BBs suggest that insulin 1 is either the exclusive or almost exclusive contributor to plasma insulin in tilapia. In the context of our transgenic fish with a humanized NTins1, it seems exceedingly unlikely that the NTins2 gene would need to be silenced.

Abbreviations

aa

amino acid

BB

Brockmann body

α (1,3)Gal

Galactose-α-1,3-galactose

NTins2

Nile tilapia insulin 2

NTins1

Nile tilapia insulin 1

PCR

polymerase chain reaction

qPCR

quantitative PCR

RACE

rapid amplification of cDNA ends

RT

reverse transcription

Disclosure of Potential Conflicts of Interest

No potential conflicts of interest were disclosed.

Funding

The authors thank the Juvenile Diabetes Research Foundation and the University of Calgary Faculty of Medicine for funding supporting this work.