Iridescence in L. pealeii skin can be induced by the exogenous addition of the neurotransmitter ACh (). Rapid changes in both the intensity and colour of reflectance are observed following the addition of ACh (). The calcium ionophore A23187 also induces iridescence in L. pealeii without the addition of ACh (b) (cf. Cooper & Hanlon 1986). The spectral shift following the initiation of iridescence via the addition of ACh was observed to proceed from 680 to 650 nm () resulting in a visible change in iridophore colour from red to orange-yellow. In addition to blue-shifting the peak of the reflectance, longer ACh stimulation resulted in a narrowing of the peak, or a decrease in the full width at half maximum of the reflectance. This optical effect is associated with an increase in the chroma, or the purity and intensity of the ACh-stimulated structural colour.
We extracted and characterized four insoluble proteins expressed only in the iridophore-containing layer of the skin (). Two of these proteins, with apparent molecular masses of 40 and 25 kDa based on SDS–PAGE mobility, strongly cross-react with anti-reflectin antibodies generated against E. scolopes reflectins. The remaining two proteins exhibit SDS–PAGE molecular masses of 30 and 20 kDa and only weakly cross-react with the anti-E. scolopes reflectin antibody (, right).
Figure 3. Electrophoresis, immunoreactivity and phosphorylation of reflectins. Centre panels: overlay of two artificially coloured two-dimensional electrophoretic blots of reflectin proteins stained with Pro-Q Diamond stain for phosphoamino acids (a) or PY-20 antibody (more ...)
We cloned and sequenced the genes coding for three of these proteins and designated them as follows: the strongly immunoreactive proteins reflectin A1 (44 kDa) and reflectin A2 (25 kDa); and the weakly immunoreactive protein reflectin B1 (30 kDa). Their deduced amino acid sequences () revealed similarity to the reflectin proteins of E. scolopes
and L. forbesi
(Crookes et al. 2004
; Weiss et al. 2005
), with reflectins A1 and A2 exhibiting greater sequence homology to the previously characterized proteins than reflectin B1. Interestingly, reflectin B1 appears to be the most abundant protein we have extracted from the dynamically iridescent tissue (). Although seven unique reflectin gene variants were identified in E. scolopes
(Crookes et al. 2004
), a similar number of reflectin sequences has not yet been found in L. pealeii
Figure 4. Sequence alignment and MALDI-TOF confirmed phosphorylation sites of novel reflectin proteins. The three novel reflectin proteins characterized here are aligned with reflectin 1B. Phosphorylated residues identified in MALDI-TOF MS analysis are highlighted (more ...)
As is the case for the other reflectin proteins, reflectins A1 and A2 contain a series of conserved reflectin motifs (RMs) interspersed throughout the protein, previously defined as [M/FD(X)5
] (Crookes et al. 2004
). There is an additional conserved N-terminal motif [MEPMSRM(T/S)MDF(H/Q)GR(Y/L)(I/M)DS(M/Q)(G/D)R(I/M)VDP(R/G)] in both Loligo
reflectins. This N-terminal region is more evolutionarily conserved across species and reflectin isoforms than the canonical RM. This highly conserved N-terminal domain aligns poorly with the previously defined repetitive motif, and is not repeated within the reflectin proteins. The N-terminal region is the only conserved motif present in reflectin B1, which does not contain any of the other canonical RMs as previously defined (). In contrast, reflectin A1 contains five canonical RMs and reflectin A2 contains three canonical RMs in addition to the N-terminal motif.
All three of the reflectins we characterized had exceptionally high aromatic amino acid residue content: 19 per cent in reflectin A1, 19 per cent in reflectin A2 and 19 per cent in reflectin B1. We have included histidine in our accounting of aromatic residues, since, for reasons discussed below, it appears to behave like an aromatic residue in reflectin proteins. In contrast, the average content of aromatic residues in most other proteins is approximately 10 per cent. The total content of arginine and methionine also is quite high in our reflectin proteins: arginine content was 12, 11 and 10 per cent for reflectins A1, A2 and B1, respectively. Methionine content was 16, 14 and 11 per cent for reflectins A1, A2 and B1, respectively. Notably, these reflectins have extremely few non-aromatic hydrophobic residues that are usually required for protein folding. Reflectin B1, although it contains no internal reflectin motifs as previously defined, shares the N-terminal motif and other properties such as high methionine content and high positive charge with previously identified reflectins ().
After complete acid hydrolysis of the dermal iridophore layer under vacuum, no glucosamine was detected by amino acid analysis, indicating the absence of chitin.
Hydropathy plots of the L. pealeii reflectins revealed that the three proteins are globally hydrophilic (data not shown), lacking any distinct hydrophobic regions. Computational predictions of trans-membrane helices were also negative. Although hydrophobicity calculations predict that reflectins should be water-soluble, upon purification they co-migrate with the cellular membrane fraction. Accordingly, plots of membrane-interface affinity showed that the non-conserved regions between reflectin motifs are highly energetically stable in membrane-interface regions (). The predicted isoelectric points for the three new reflectin proteins are >8.5; these high isoelectric points are largely due to high arginine content, indicating that these proteins are likely to be positively charged in the absence of extensive phosphorylation and under physiological conditions.
Figure 5. Interface affinity of reflectin proteins. Traces indicate the calculated energy of interface interaction between a hypothetical membrane surface and 13-residue ‘windows’ of the reflectin proteins as a function of position in the proteins. (more ...)
Using neural network predictions (Blom et al. 1999
), we predicted numerous potential sites of phosphorylation for the three Loligo
proteins. For reflectin A1, we identified 10 serine (Ser), 13 tyrosine (Tyr) and zero threonine (Thr) sites; for reflectin A2, we identified five Ser, 11 Tyr and two Thr sites; for reflectin B2, we found 20 Ser, six Tyr and one Thr sites. Analysis of the all three reflectin sequences using PROSITE (Hulo et al. 2006
) and Pfam (Finn et al. 2008
) predicted no calcium-ion binding motifs (such as EF-hand motifs).
We tested the effects of several tyrosine kinase and protein kinase C (PKC) inhibitors on ACh-induced iridescence in Loligo. Genistein, a broad-range tyrosine kinase inhibitor, dramatically suppressed ACh-induced dynamic iridescence, while PKC-specific inhibitors did not. The effect of genistein on iridescence intensity exhibited a clear dose-dependency: higher doses of genistein resulted in progressively dimmer iridocyte reflectance induced by ACh ().
We confirmed and quantified active phosphorylation of the three Loligo reflectins by staining SDS–PAGE separated reflectins with Pro-Q Diamond, a stain specific for phosphoamino acids. We compared staining intensities of ACh-treated tissue with those in the ACh + genistein-treated tissue. These intensities were normalized to those of untreated control tissue. We found that both reflectin A1 and reflectin A2 showed higher phosphorylation levels in the ACh-only treatment () than in either the ACh + genistein treatment or the control treatment. The ratio of normalized staining intensity between the ACh-only and the ACh + genistein-treated tissues increased by 130 and 240 per cent for reflectins A1 and A2, respectively (). In contrast, the net phosphorylation staining intensity of reflectin B2 decreased by 25 per cent ().
Figure 6. Changes in phosphorylation with ACh treatment and genistein + ACh treatment of iridophores. Left, changes in Pro-Q Diamond staining intensity of reflectin proteins upon treatment with 10 µM ACh (red columns) or 10 µM ACh + genistein at (more ...)
Since Pro-Q Diamond cannot differentiate between different phosphorylated amino acids, we specifically quantified levels of tyrosine phosphorylation via Western blotting with PY-20, a phosphotyrosine-specific antibody. Consistent with the results from Pro-Q staining, phosphotyrosine levels for the ACh-treated tissue increased by 170 and 290 per cent for reflectins A1 and A2, respectively, when compared with the ACh + genistein treatment (). The control values are included in the data presented in , as the data for the experimental treatments are plotted as the per cent change relative to the control values obtained from the corresponding (control) lane of the gel. We were unable to detect reflectin B1 with PY-20 in any tissue or treatment, indicating that tyrosine phosphorylation is low, not present, or especially labile for this protein. PY-20 sensitivity is known to be dependent on sequence context, possibly contributing to this observation. Alternatively, this result may suggest that phosphorylation of serine and/or threonine dominates in the active state for reflectin B1, in contrast to the phosphorylation of tyrosine residues in the reflectin A sequences.
To identify the locations and identities of the reflectin residues phosphorylated in the ACh-activated tissue, MALDI-TOF MS was performed on trypsin digests of reflectins A1 and reflectin A2 (). This technique revealed that Tyr14 and Tyr127 of reflectin A1 and Tyr12, Tyr 214, Ser218 and Tyr223 of reflectin A2 were phosphorylated in the active state (), consistent with our stain-based analysis of phosphorylation. Reflectin A1 had an additional peptide (WMDAQGRFNNQFGQMWHGR) that showed a mass shift consistent with phosphorylation despite the absence of tyrosine, serine or threonine. This may indicate a histidine phosphorylation, as indicated in . In those cases in which more than one phosphorylatable amino acid was present in a mass-shifted peptide, we used the congruence of our computational phosphorylation predictions and our MS data to identify the specific residues that were phosphorylated. Interestingly, all but one phoshorylation event occurred outside the conserved reflectin domains. It is possible that additional non-tyrosine phosphoamino acids may have been missed by our analyses because of their lability during protein isolation and purification (Sickmann & Meyer 2001
; Mikesh et al. 2006
Reflectin phosphorylations predicted from MALDI-TOF MS. Peptide sequences identified by MALDI-TOF MS, with asterisks indicating identified phosphorylated residues.
Two-dimensional PAGE revealed the populations of phosphorylated states of the reflectins before and after iridescence activation (). The multiple phosphorylated states of each isoform reflect the fact that each reflectin contains multiple residues that can be phosphorlyated. However, because reflectin A1 has very limited solubility in the IEF buffer in the absence of detergent (as required for two-dimensional PAGE), and the different reflectin isoforms exhibit variable solubilities in detergent-free buffer, it proved difficult to quantitatively compare phosphorylation levels between reflectin species in the same two-dimensional PAGE experiment. In contrast, we found that reproducible comparisons of the levels of phosphorylation can be obtained from the analyses of one-dimensional gel electrophoresis conducted in the presence of SDS. Our two-dimensional PAGE data enabled us to evaluate the shifts in the populations of differentially phosphorylated states of the reflectins from untreated and ACh-treated samples. Thus, staining with Pro-Q Diamond and PY-20 revealed that reflectin A1 and reflectin A2 each consist of populations with several discrete phosphorylated states for both activated and inactivated tissues (). Immunodetection with PY-20 (and, to a lesser degree, staining with Pro-Q Diamond) revealed that the most acidic, extensively phosphorylated reflectin A2 molecules possess an isoelectric point of 7, and are in significantly higher abundance in the activated tissue. Reflectin B2 also appeared to exhibit a higher degree of phosphorylation, as determined with Pro-Q Diamond, following the addition of ACh (). In contrast, Pro-Q Diamond staining shows that reflectin B1 became more basic (less phosphorylated) upon activation, and also is present in several distinct phosphorylated populations (). In general, the reflectin proteins extracted from the ACh-treated tissues appeared to migrate slightly farther in the first dimension of two-dimensional PAGE than did the proteins from the controls. Although we cannot rule out imperfect alignment of gel images, it seems probable that this is caused by increased negative charge on the activated proteins due to phosphorylation.