OUP user menu

Retinoic acid as cause of cell proliferation or cell growth inhibition depending on activation of one of two different nuclear receptors

George Wolf
DOI: http://dx.doi.org/10.1111/j.1753-4887.2007.00006.x 55-59 First published online: 1 January 2008


Retinoic acid can combine with the nuclear retinoic acid receptor (RAR), leading to cell growth inhibition, as in certain tumors. Retinoic acid can also bind to the orphan nuclear receptor peroxisome proliferator-activated receptor β/δ (PPARβ/δ), resulting in stimulation of cell growth and inhibition of apoptosis. To bind to RAR, retinoic acid is carried into the nucleus by the cytosolic cellular retinoic acid-binding protein-II; to bind to PPARβ/δ, it is transported into the nucleus by the cytosolic fatty acid-binding protein 5.

  • cellular retinoic acid-binding protein-II
  • fatty acid-binding protein 5
  • keratinocytes
  • mammary carcinoma MCF-7
  • peroxisome proliferator-activated receptor β/δ
  • retinoic acid receptor


Textbooks on nutrition broadly describe vitamin A to function as a cofactor in the process of vision, as a hormone in reproduction, in embryonic development, bone growth, the immune system, the growth and differentiation of epithelia, and the inhibition of tumor growth. Clearly, stimulation of growth and differentiation of epithelia would have to occur by mechanisms different from inhibition of the growth of tumors. A recent report by Schug et al.1 describes mechanisms whereby two such opposing activities can be performed by one vitamin.

Vitamin A, known as retinol, functions in the form of its metabolite, retinoic acid (RA), by binding to, and thus activating, its cognate nuclear hormone receptors RARα, β, and γ. In combination with the activated retinoic acid X receptor (RXR), a heterodimer is formed (RAR-RXR), which binds to specific target genes, leading to their transcriptional expression, and ultimately to the regulation of their function.

The two opposing mechanisms taken as examples by Schug et al.1 were 1) the stimulation by RA of proliferation of the basal keratinocytes of the skin and 2) the well-known growth inhibition of cancer cells by RA. The first clues to a reconciliation of these opposing effects were obtained by Chapellier et al.2 It appeared that in the skin of mice in which the normal RA receptors RARα, β, and γ were knocked out, RA-induced keratinocyte hyperplasia still took place. The authors2 found that the activation by RA of a member of an entirely different family of nuclear receptors, the orphan receptor PPARβ/δ, resulted in enhanced mitosis of cells of the basal layer of the keratinocytes of the skin, maintaining skin integrity and the epidermal barrier after wounding.3 To follow up this finding, Shaw et al.4 reported that, in fact, RA binds with high affinity to and activates PPARβ/δ. Fluorescence titration showed that PPARβ/δ bound RA with KD = 17 nM, whereas PPARα and γ bound RA with KD = 100–200 nM. They found that, together with the co-activator SRC-1, binding of RA to PPARβ/δ led to upregulation of the expression of 3-phosphoinositide-dependent kinase 1 (PDK1), a factor downstream of PPARγ/δ, and subsequently the activation of the anti-apoptotic factor Akt1.

The authors4 thus were led to “the surprising conclusion” that RA can stimulate cell proliferation through binding to PPARβ/δ and, as is well known, inhibit cell growth, as in cancer cells, by binding to RAR. The following question then arose: How can RA be partitioned between these two nuclear receptors, leading to two outcomes?


Budhu and Noy,5 in earlier work, found that the well-known intracellular cytosolic RA-binding protein CRABP-II “massively” transported RA into the nucleus. There, the RA became attached to RAR. The resulting CRABP-II-RA-RAR complex was short-lived, releasing CRABP-II and allowing the RA-RAR to perform its function of interacting with the appropriate gene. This directed the cell towards growth inhibition. In confirmation, the authors5 showed that overexpression of CRABP-II in MCF-7 mammary carcinoma cells greatly increased the growth inhibitory action of RA; the reduced expression of CRABP-II in these cells made them insensitive to RA. Thus, one channel exists for directing RA to RAR, leading to growth inhibition.

The earlier work of Shaw et al.4 determined that, as described above, RA can combine with and activate PPARβ/δ. Could a second channel exist that leads to the opposite effect of RA action, namely induction of growth stimulation, as found in keratinocytes? Indeed, Tan et al.6 showed that a particular fatty acid-binding protein, FABP5, “massively” transported RA from the cytosol to the nucleus, where it activated PPARβ/δ.

In the most recent report from Noy's laboratory1 the following hypotheses were proposed and tested: 1) that RA stimulates cell proliferation (as in keratinocytes) by binding to the nuclear receptor PPARβ/δ, channeled to the nucleus by a fatty acid-binding protein, FABP5; and 2) that RA inhibits cell growth (as in MCF-7 carcinoma) by binding to the nuclear receptor RAR, channeled into the nucleus by CRABP-II.


The authors1 employed the following two cell types, which respond to RA by stimulation of growth: 1) the human keratinocyte cell line HaCaT; and 2) NaF cells derived from a mouse mammary cancer model TgN (MMTVneu) 202Mul. They showed that RA uniquely stimulated growth of these tumors in mice that bore them.

In order to study the action of RA on HaCaT keratinocytes, they first determined the specificity of the PPARβ/δ response. They showed that a specific synthetic PPARβ/δ-selective ligand, GW0742, activated a luciferase reporter driven by a PPAR response element on the DNA, whereas a RAR-selective ligand, TNPB, was inactive. They then treated the cells with RA and found a dose-dependent response of the PPAR-driven luciferase. Furthermore, this activation of PPARβ/δ was suppressed in the presence of the inhibitory siRNA directed at PPARβ/δ. The important conclusion from these experiments was that RA activated PPARβ/δ in the keratinocytes.

Next, to confirm this conclusion, the authors1 investigated the effect of RA treatment on the endogenous target genes of PPARβ/δ in the HaCaT cells. As a positive control, they again showed that the specific PPARβ/δ activator GW0742 stimulated the expression of the following three downstream target genes: PDK1, fasting-induced adipose factor, and adipose differentiation-related factor. They then showed that RA treatment greatly upregulated the expression of all three genes. The RA isomer 9-cis-RA had a very small effect.

In keratinocytes, the pro-proliferation (growth-stimulating) activity of RA is accompanied by anti-apoptosis action. The anti-apoptosis gene activation leads to phosphorylation and thereby activation of the downstream factor Akt. The investigators1 found first that treatment with GW0742, the specific activator of PPARβ/δ, increased phosphorylation of Akt. They then showed that RA had the same effect on Akt. This was evidence that RA, through activation of PPARβ/δ, as well as being a proliferation stimulator, worked as an anti-apoptosis agent.

To establish these conclusions more firmly, the authors1 showed, by fluorescence titration, that both RA and GW0742, the specific synthetic ligand for PPARβ/δ, readily bound to PPARβ/δ with KD = 42.3 ± 4.5 nM and KD = 34.8 ± 6.6 nM. Also, in a control experiment, COS-7 cells transfected with FABP5 fused to green-fluorescent protein, showed distribution of the green fluorescence all over the cells when treated with stearic acid. On the other hand, when treated with RA, the fluorescence shifted totally into the nucleus, showing that the FABP5 present combined with the RA and migrated to the nucleus. Similarly, in the human keratinocytes HaCaT, addition of RA (but not stearic acid) led to the accumulation of FABP5 in the nucleus, as revealed by subcellular fractionation. Thus, the channeling of RA into the nucleus, to meet and activate PPARβ/δ, was firmly established.

Parallel to the activation of PPARβ/δ by RA, this vitamin metabolite also activated RAR, its traditional nuclear receptor in the same keratinocytes. Thus, the authors1 found that RA in the keratinocytes upregulated twofold a reporter gene driven by an RAR construct. Clearly, even in keratinocytes, which respond to RA by proliferation through PPARβ/δ activation, a pathway exists for activation of RAR.


Earlier work from the same laboratory5,7 had established that, when RA binds to its intracellular binding-protein CRABP-II, RA is transported into the nucleus. There, the complex RA-CRABP-II attaches to and activates RAR while CRABP-II is released. The authors1 next provided evidence that RA can activate PPARβ/δ or RAR, depending on binding to either FABP5 or CRABP-II. COS-7 cells were transfected with a luciferase reporter construct driven by a PPAR-response element (PPARE) expression vector for PPARβ/δ, plus a vector for either FABP5 or CRABP-II. When treated with RA, the PPARE reporter was activated in a dose-dependent manner. The presence of the FABP5 vector greatly enhanced RA-mediated PPARβ/δ transactivation. The presence of the CRABP-II vector had no effect on the PPARE reporter.

The converse experiment was done with a luciferase reporter driven by a RAR response element, a construct for RARα and vectors for either FABP5 or CRABP-II. The channeling of RA through CRABP-II to RARα resulted in the reporter responding dose-dependently, whereas the cells harboring FABP5 were not affected. These experiments showed that cells artificially provided with the appropriate expression vectors were able to channel RA to the nucleus by the fatty acid-binding protein FABP5 to activate PPARβ/δ, and by the cellular retinoic acid-binding protein CRABP-II to activate RARα.

Finally, HaCaT keratinocytes were transfected with FABP5-siRNA, causing an 80% reduction in FABP5. This procedure greatly lowered the ability of either GW0742 (the specific activator of PPARβ/γ) or RA to stimulate expression of ADBP, a downstream factor of PPARβ/δ activity. This experiment showed that FABP5 is necessary for the RA effect in activating PPARβ/δ in the human keratinocyte cell line.


So far, the investigators1 had established that RA is transported to one of its nuclear receptors, PPARβ/δ by FABP5 and is shuttled to the other, RAR, by CRABP-II. The result was that the first process was growth stimulating, the second was growth inhibiting. So for instance, in MMTVneu mammary tumor cells, as shown by immunoblots,1 the level of FABP5 was found to be high and that of CRABP-II to be low; in adjacent normal tissue, CRABP-II was high and FABP5 was low. Cancer cells that respond to RA by growth inhibition showed a high ratio of CRABP-II/FABP5. In confirmation, Manor et al.8 showed that ectopic expression of CRABP-II in mammary carcinoma SC115 cells greatly enhanced the tumor's sensitivity to RA-induced growth inhibition.

The authors1 then asked, what would happen if the ratio FABP5/CRABP-II in cells were to be artificially inversed? Taking the extreme case of keratinocytes HaCaT, containing a high level of FABP5 and almost undetectable CRABP-II, the authors1 investigated the reversal of the FABP5/CRABP-II ratio in two different ways: 1) by overexpressing CRABP-II in the keratinocytes; and 2) by transfection of siRNA for FABP5 that suppressed FABP5 expression. The level of expression of PDK1, the downstream target of PPARβ/δ was assayed. The results were clear: 1) raising the CRABP-II/FABP5 ratio by over-expression of CRABP-II abolished the effectiveness of RA to induce PDK1; 2) under-expression of FABP5 abolished the effectiveness of RA to induce PDK1; 3) under-expression of FABP5 enhanced the stimulation of PDK1 by GW0742, the synthetic ligand for PPARβ/δ.

These results demonstrated that increasing the CRABP-II/FABP5 ratio in the keratinocytes directs the RA from PPARβ/δ and towards inhibitory action on cell proliferation through decreased action on PPARβ/δ and hence PDK1. The results also showed that FABP5 is a carrier of RA into the nucleus, where it can activate PPARβ/δ.

In view of the high FABP5/CRABP-II ratio in the keratinocytes, one would expect that in these cells RA would not exert the traditional pro-apoptotic action that RA effects in cancer cells, such as in the MCF-7 carcinoma. In order to explore the effect of inversing the FABP5/CRABP-II ratio on apoptosis in keratinocytes, the authors1 treated the cells with the apoptosis-inducing tumor necrosis factor-α (TNF-α). They assayed the effect by determining the specific cleavage of ADP-ribose polymerase (PARP), one of the earliest events in apoptosis.

Earlier work9 had shown that PPARβ/δ protected keratinocyes from apoptosis. In the present work1 the authors found that in keratinocytes the activation of PPARβ/δ by the synthetic ligand GW0742 inhibited TNF-α-induced apoptosis. Furthermore, RA protected the cells from apoptosis, presumably by combination with and activation of PPARβ/δ. On the other hand, the synthetic RAR ligand TTNPB induced apoptosis in the keratinocytes by itself and enhanced this effect when given together with TNF-α, no doubt by its activation of RAR. Overexpression of CRABP-II in the keratinocytes caused the loss of the apoptosis-protective action of RA, but acted like TTNPB in inducing apoptosis, as a result of the change in the CRABP-II/FABP5 ratio.

As outlined above, NaF cells, derived from the mouse mammary carcinoma MMTVneu, which the authors1 had found to be growth-stimulated by RA, rather like keratinocytes, behaved similarly to HaCaT cells: RA prevented them from TNF-α-induced apoptosis; and both over-expression of CRABP-II or under-expression of FABP5 increased the apoptotic response.

The important conclusion reached by the authors1 was thus “in HaCaT cells, RA inhibits apoptosis like a bona-fide PPARβ/δ ligand, an activity that diametrically contrasts with the pro-apoptotic activities of RAR”. Thus, the authors1 had established the effect of artificially reversing the naturally occurring very low ratio CRABP-II/FABP5 on apoptosis in keratinocytes and NaF cells, shown to respond to RA treatment by proliferation and inhibition of apoptosis. They next turned to a cell line with a naturally high ratio CRABP-II/FABP. They chose the mammary carcinoma cell line MCF-7, responding to RA treatment by growth inhibition and apoptosis.

The authors1 investigated the effect of artificially reversing the CRABP-II/FABP5 ratio on the RA response of these cells. The ratio was reversed by co-transfection with an expression vector for FABP5 together with a construct of siRNA for CRABP-II. The resulting cell line showed increased FABP5 and decreased CRABP-II levels. The expression of the BTG2 gene following treatment of these cells with RA was measured. This gene normally causes cell-cycle arrest through activation of the RAR receptor8 and thereby stops cell proliferation. In these cells with a reversed CRABP-II/FABP5 ratio, the action of RA decreased the expression of BTG2 compared to control cells with the normal ratio.

In addition, it was found that the artificially induced increase in FABP5 and decrease in CRABP-II in the MCF-7 mammary carcinoma cells directed RA to PPARβ/δ, as revealed by an increase in the expression of the survival factor PDK1, a target for PPARβ/δ.

To gild the lily, the authors1 studied the response of the MCF-7 cells with an artificially high ratio of FABP5/CRABP-II to treatment with an apoptosis-inducing agent called tumor necrosis factor-related apoptosis-inducing ligand (TRAIL). The cells, when treated with RA in conjunction with TRAIL, showed inhibition of the normal TRAIL-induced apoptosis. In other words, normal inhibition of cell proliferation in this cell line by RA through apoptosis was reversed into “pro-survival” through reversing their normal high CRABP-II/FABP5 ratio.

In summary, the important discovery described here shows that RA can activate the transcription of two different nuclear receptors, RAR and PPARβ/δ. The activation of one, RAR, which is by far the commonest, can give rise to inhibition of cell growth. The activation of the other, PPARβ/δ, leads to expression of genes producing cell proliferation. Since both nuclear receptors can occur in one and the same cell, there exists a partitioning mechanism, whereby RA is transported from the cytosol to the nucleus by two different transporting proteins. The activation of RAR occurs when RA is delivered to RAR by cytosolic CRABP-II; the activation of PPARβ/δ occurs when RA is delivered to PPARβ/δ by FABP5. Therefore, in cells with a high cytosolic CRABP-II/FABP5 ratio, RA functions by being delivered to RAR by CRABP-II, leading to growth inhibition. In cells with a high cytosolic FABP5/CRABP-II ratio, such as keratinocytes, RA functions by being carried by FABP5 to its cognate nuclear receptor, PPARβ/δ, resulting in growth stimulation (Figure 1).

Figure 1

A model outlining the dual transcriptional activity of RA. CRABP-II and FABP5 target RA to RAR and PPARβ/δ, respectively. In cells that express a high CRABP-II/FABP5 ratio, RA is channeled to RAR, often resulting in growth inhibition. Conversely, in the presence of a low CRABP-II/FABP5 expression ratio, RA is targeted to PPARβ/δ, thereby upregulating survival pathways.

Reprinted from Schug et al.1 (Cell 2007;129:723–733) with permission.

The authors1 emphasize that the binding affinities of RA for CRABP-II and RAR are high (KD in the 0.1–0.2 nM range), whereas affinity of RA is much lower for FABP5 and PPARβ/δ (KD in the 10–50 nM range). Therefore, in most cells, RA would bind predominantly to RAR.

One of the earliest signs of a deficiency of vitamin A (apart from night blindness) is the degeneration of epithelia. One could suppose that RA stimulates growth of basal epithelial cells. This would require a high FABP5/CRABP-II ratio and a relatively high concentration of RA. A determination of RA in tissues was recently carried out by Kane et al.10 and showed a concentration of 7–9.6 pmol/g tissue in liver, kidney, testis, and brain. Epithelial concentration was not measured, but would be close to the 10–50 nM level required for interaction with FABP5, if it were similar to the concentrations found in other tissues.

The impressive work by Schug et al.1 with its compelling evidence leaves open a number of questions. The FABP5/CRABP-II ratio in epithelia generally was not determined in this work, and the keratinocytes in which this ratio was shown to be high, may represent a special case of epithelia. Also, as stated by the authors, the fact that a high CRABP-II/FABP5 ratio is much commoner in tissues would imply that RA acts generally as a cell growth inhibitor. This is the case for certain tumors; however, the nuclear RAR receptors activated by RA appear to have many other functions.


View Abstract