To analyze the relative amounts of two differently phosphorylated MARCKS variants present in the same immunoreactive electrophoretic band, we introduced here a new and simple procedure for processing Western-blot images. Merging digitally colored immunoreactive bands permitted us to obtain a quick semi-quantitative evaluation of polypeptide amounts. This procedure was used to compare the quantity of total MARCKS, pED-MARCKS and S25p-MARCKS during neural retina development, as well as in several different experimental situations. Therefore, it was possible to show that MARCKS was present since the earliest neural retina developmental stage assayed here, and that the same holds for S25p-MARCKS (initially expressed by differentiating retinal ganglion cells), confirming our earlier findings 
. In addition we showed the presence of pED-MARCKS in vivo
already at E4, although only the mitotic cells were immunoreactive. It must be remembered that at this developmental stage there is an important cell proliferation activity in the neural retina, which explains why this isoform constitutes a large fraction of the total protein. pED-MARCKS cell distribution has recently been described in dividing human hepatic stellate cells, where MARCKS was highly phosphorylated and localized at distinct sub-cellular regions in each one of the mitotic phases 
. Developmental changes in the amount of pED-MARCKS have also been found in rat brain, where this MARCKS phosphorylated variant was not detectable in embryos before the 14th day, but increased at birth reaching the highest values at P5–P22 
. S25p- and pED-MARCKS displayed very different spatial and temporal expression patterns during neural retina development. While the former accumulated with the progression of cell differentiation and spanned the whole retinal width, the latter appeared to vary randomly and was mainly confined to the retinal ganglion cells and their axons as well as to mitotic cells.
Taking into account the comparison of MARCKS protein sequences from several vertebrate species, a high homology was found in the region named “MH2”, S25 to E49 in the chick sequence (S26-E50 in humans) 
. Part of this conserved protein domain was initially identified as the one comprising the RNA splicing region, which could be the reason for its high level of preservation 
. The fact that a phosphorylation site specific for differentiating neurons is highly conserved among all MARCKS sequences identified so far, also suggests that it might be important for its function in those cells. In addition, S25 is inserted in a consensus sequence for Cdk phosphorylation (VAASPSK) recognized by ELM (“Eukaryotic Linear Motif”; 
) web site as such; and as we mentioned in the Introduction, human MARCKS Serine 26 was phosphorylated by Cdk2 in in vitro
Our results from kinase inhibition experiments indicate that MARCKS phosphorylation at S25 is maintained by a permanent activity of a Cdk, which must be counteracted in cells by some unidentified phosphatase(s). As cell-cycle related Cdks are not active in differentiating retinal neurons 
, we deduced that the kinase responsible for this modification is Cdk5. Cdk5 is a multiple-target Ser/Thr protein kinase that has been involved in several processes during neuronal differentiation 
. Many of its known targets are cytoskeleton-modulating proteins, such as Tau and neurofilament proteins, as well as some intracellular signaling proteins. In the case of MARCKS, previous work from Yamamoto and collaborators has shown that Cdk5, as well as Cdk1, are able to phosphorylate some unidentified rat MARCKS serine and threonine residues in vitro
. However, kinase inhibitors are usually not completely specific, and it has also been shown that there is some extent of interaction of the activity of kinases like Cdk5 and GSK3β, and that they may even phosphorylate some proteins at the same sites 
. Results from pharmacological experiments shown here demonstrate that GSK3β does not phosphorylate MARCKS at S25 in a detectable way in living cells. Interestingly, the same experiments also indicate that MARCKS phosphorylation by this enzyme would not be necessary for priming the one produced on S25 by Cdk5. This concerted action of both enzymes is a known phenomenon, occurring on Tau, where Cdk5 modulates some GSK3β specific site phosphorylations 
. The pharmacological inhibition experiments presented here also indicate that Cdk5 would be the site-specific kinase phosphorylating MARCKS, not only at the onset but also at the progression, of retinal neuroblasts differentiation. On the other hand, this same enzyme produces the phosphorylation of MARCKS in cells after a near complete dephosphorylation of S25, when actin filaments are disassembled, as described 
Additional information to reinforce the idea that Cdk5 does indeed interact with MARCKS in vivo
, derives from the observation that both proteins are found accumulated in the same cells in the retina, and partially co-localize in growing neurites and growth cones. Their incomplete subcellular co-localization is most probably related to the fact that, in differentiating neurons, Cdk5 is associated to the cytoskeleton, where it phosphorylates many other substrates 
. As we found both proteins associated to low-density membrane microdomains (Fig. S2
), it is tempting to speculate that these complexes are some type of actin-associated membrane rafts. p35 and p39, Cdk5-activating subunits, are myristoylated proteins 
, hence active Cdk5 could in theory interact at the plasma membrane with the same microdomains as MARCKS. But we cannot rule out other hypothesis: since phosphorylation at S25 appears to be very stable it could also be possible that MARCKS phosphorylation by Cdk5 occurs at a cellular compartment apart to where the protein is actually functional.
We previously showed that the persistence of S25 phosphorylation depends on actin filaments stability 
. Now we describe that the phosphorylation at MARCKS ED by PKC provokes a rapid S25 dephosphorylation. The attenuation, and even the disappearance of this phosphorylation, occurs as a function of PMA concentration and treatment duration. Moreover, it is enough to stimulate PKC with PMA during one minute to produce two effects: a) a progressive increase of ED phosphorylation, and b) an associated decrease of S25 phosphorylation. Our results concerning ED phosphorylation differ from the early ones obtained by Rozengurt stimulating Swiss 3T3 intact cells with phorbol 12–13 dibutyrate (PBt2
) for 2 minutes. In these cells, the 80 kDa protein (MARCKS) was rapidly phosphorylated to a maximum followed by a decay attaining minimal values 6 minutes after PBt2
. As far as we know our results are the first reporting the existence of a relationship between two MARCKS phosphorylated sites, produced by different kinases. It was clear from our results that there is a decay of S25p-MARCKS when Cdk5 is inhibited in intact cells. This decay would be indicating that the involved specific phosphatase is active, but that the phosphorylation is not sustained as a consequence of the kinase inhibition by roscovitine or olomoucin. Interestingly, in conditions of PKC stimulation by PMA, we observed a faster loss of S25 phosphate. This result strongly suggested that PKC would be activating the phosphatase and/or perhaps modifying the kinetics of Cdk5.
It is known from the results of Yamamoto 
, that PP2A is the phosphatase removing the MARCKS phosphates incorporated by proline-directed kinases, like Cdks. An analogous phenomenon was found at the dopamine regulation of DARPP-32 phosphorylation, which occurs at its threonine 75 in neostriatal neurons. In this case it is the cAMP dependent protein kinase (PKA) that, once stimulated by dopamine, increases PP2A activity without modifying the activity of Cdk5; at the same time it phosphorylates DARPP-32 at another residue, threonine 34 
. More recently it was reported that the activation of PP2A is due to a phosphorylation of its B56d subunit, produced by PKA 
. Later on, these authors found that stimulated PKC is able to phosphorylate B56d subunit, leading to an activation of PP2A, which produced the dephosphorylation of Ser 40 of tyrosine hydroxylase 
Most of the previous research regarding MARCKS serine dephosphorylation performed either using the purified protein or on different cell types, has been devoted to those residues located at the ED and phosphorylated by PKC. ED dephosphorylation has been attributed to phosphatases PP1A, PP2A, PP2B and PP2C 
. Nevertheless, MARCKS has several other serine and threonine residues known to be phosphorylated by different protein kinases (see review by Mosevitsky 
), although only a few experiments explored their dephosphorylation in vitro
. In fact, Yamamoto et al. 
showed that MARCKS, phosphorylated by PKC, cdc2 or Cdk5, was only dephosphorylated by the holoenzyme of PP2A and not by PP2B. We assayed the effect of various protein phosphatase inhibitors in several treatments applied to intact neuroblasts. The results presented here show that only calyculin A attenuates the dephosphorylation induced by PMA treatment. All the drugs had clear inhibitory effects over phosphatases acting on the phosphorylated MARCKS ED or, in the case of okadaic acid and calyculin A, on GSK3β phospho-serine 9, indicating that in neuroblasts both substrates are sensitive to phosphatase inhibition with okadaic acid. It is of note that in other cell types such as Swiss 3T3, okadaic acid inhibits phosphatase 2A acting on MARCKS ED at concentrations that are extremely toxic for neuroblasts 
From our experiments, PP1 and PP2B can be excluded as putative phosphatases for MARCKS serine 25, given the non-inhibitory effects of tautomycetin (PP1), FK506 and cyclosporine A (PP2B). Experiments using calyculin A as protein phosphatase inhibitor pointed to a form of PP2A as the most probable phosphatase involved in MARCKS S25 dephosphorylation. It is known that this enzyme is highly versatile depending of its regulatory subunit (reviewed in 
). It has been demonstrated that PKC phosphorylation of a PP2A subunit causes an increase of its catalytic activity, as described by Ahn et al. 
for tyrosine hydroxylase dephosphorylation. This would be considered a potential explanatory mechanism for MARCKS dephosphorylation under our experimental conditions.
The cellular phenomenon described here appears singularly complex, particularly in which concerns to the molecular mechanisms that would involve structural aspects as well as regulatory networks. In fact, it is necessary to elucidate the dynamic structural properties of the MARCKS region in which S25 lies, and that could account for the peculiarities of its phosphorylation/dephosphorylation cycle in cells. At the same time it is necessary to gain a better understanding of many intricate signal transduction networks acting during neuroblast differentiation.