healthmedicinet

Once melanoma progresses to the invasive and metastatic stage, it is very difficult
to treat. Therefore, it is important to identify the molecular changes that contribute
to the malignant progression of this disease. Hypoxia and acquisition of a vascular
network together with reprograming of the cancer cell’s metabolism have been noted
as important events required for tumor progression 38], 17]. Accumulation of HIF-1? and HIF-2? was measured via immunohistochemistry in 46 patient
samples of nodular cutaneous malignant melanomas 19]. Expression of HIF-1? and HIF-2? was directly correlated with vascular endothelial
growth factor accumulation (VEGF) and also associated with poor prognosis. A later
study of 89 patients with primary cutaneous melanoma did not show a correlation between

HIF-1? and overall survival or disease-free survival 20]. However, the relative amount of HIF-1? and more importantly the activity of HIF
as assessed by target gene expression in the samples was not assayed. HIF-1? was found
under normoxic conditions in malignant melanoma cells, but not in normal human melanocytes.
Further, the amount of HIF-1? was increased in cells from invasive and metastatic
human melanomas relative to that found in cells from radial growth phase melanomas.
Also knockdown of HIF-1? in the metastatic cells led to marked decrease in anchorage-independent
growth and the ability to invade through Matrigel 25]. Kuphal et al. 18] verified the constitutive expression of HIF-1? in malignant melanoma and their studies
implicate ROS and the NF-?B pathways in contributing to this accumulation of HIF-1?.

Due to their role in tumor progression, HIF-1? and HIF are targets for the development
of new small molecular inhibitors. However, most of the inhibitors to date work in
an indirect fashion such as Bortezomib, a proteasome inhibitor and geldanamycin a
Hsp90 inhibitor 39] In fact, medicinal chemists have deemed that HIF is undrugable 39].

AA (vitamin C) plays a direct role in regulating both the activity of PHD and thus
the stability of HIF-1? and the activity of FIH, which inhibits the transcriptional
activity of HIF. There are several reports that addition of AA to cancer cell lines
decreases the amount of HIF-1? protein and also inhibits HIF activity 40]–42]. Thus AA might be useful as a direct inhibitor of the HIF pathway presumably through
its action on the family of Fe (II)-2-oxoglutarate-dependent oxygenases, of which
PHD and FIH are members. We investigated this possibility in human malignant melanoma
cells.

In agreement with the study of Knowles, et al. 40] we found that AA decreased the amount of HIF-1? protein in malignant melanoma cells
grown under either normoxic or hypoxic-mimetic (CoCl
₂
) conditions (Figs. 1, 2 and 3). Further, we showed that A2P, a less oxidizable analog of AA, was more potent than
AA in reducing the amount of HIF-1? in the melanoma cells treated with CoCl
₂
(Figs. 2 and 3). We could not find any other reports on the effect of A2P on HIF-1? levels, but
several studies show that A2P inhibits tumor invasion 43], 44], while it also inhibits melanogenesis in melanocytes 45]. A2P was also more potent than AA in reducing HIF reporter gene activity (Fig. 4 panels B C). Also note that the ability of AA to inhibit HIF reporter gene activity
was more potent than its ability to decrease HIF-1? protein levels (compare Fig. 4B with Fig. 3). This finding agrees with the report of Kuiper et al. 41] that AA preferentially suppresses the HIF-1 transcriptional response. The authors
suggest that this preference is likely due to AA’s ability to stabilize and reduce
the iron atom in the PHD and FIH active sites, with FIH (asparagine hydroxylase) being
more sensitive to fluctuations in intracellular ascorbate.

Next we investigated the hypothesis that the effect of AA on decreasing HIF-1? in
human melanoma cells was mediated through stimulation of PHD and or FIH activity.
First we used a pharmacological inhibitor of PHD and likely FIH, ethyl-3, 4-dihydroxybenzoate
(EDHB). This inhibitor decreases prolyl hydroxylase activity 46] through both competition for the AA binding site 47] and inducing an iron deficiency state in cells through a low affinity for ferric
iron 48]. In our melanoma cells, EDBH at or above a concentration of 750 ?M stimulated HIF
reporter gene activity by 4-fold. Since AA pretreatment was more effective in blocking
the EDHB stimulation of HIF-reporter gene activity than when EDHB and AA were add
simultaneously to the cells, we suggest that there is competition for either the ferric
iron or the AA binding site on the PHD/FIH enzymes (Fig. 5).

After defining the condition for maximum inhibition of PHD by EDBH as measured by
HIF-reported gene activity, we then measured the concentration-dependent ability of
pre treatment with either AA or A2P to reverse the inhibition as determined by a decrease
in HIF-reporter gene activity (Fig 6). In contrast to the hypoxia mimetic CoCl
₂
stimulation of HIF-reporter gene activity where A2P was more potent than AA in reversing
this stimulation, AA and A2P were similar in their potency for reversing EDHB stimulation
of HIF-reporter gene activity. This may be due to off-target (non-PHD) effects of
CoCl
₂
causing greater stimulation of HIF-reporter gene activity relative to the activity
induced by the PHD selective inhibitor, EDHB which may be more sensitive to AA supplementation.

Since chemical inhibitors can have off-target effects, we used siRNA to knock down
the expression of the PHD2 isoform protein. An RT-PCR survey of the expression of
PHD isozymes in the WM9 cells revealed that these cells express predominantly PHD2
and a small amount of PHD1, but we could not detect expression of PHD3. The PHD2 isozyme
contributes the majority of the HIF-hydroxylase activity in cells with normal oxygen
levels 49], 50]. Since PHD1 is localized exclusively in the nucleus 51], it should only be able hydroxylate HIF-1? after it has been stabilized and transported
into the nucleus. We were able to knockdown the expression of PHD2 in normoxic WM9
cells by greater than 90 %. As shown in Fig. 7, this knockdown resulted in a 2.3 fold higher amount of HIF-1? relative to cells
treated with the control siRNA. Despite the knockdown of PHD2 and the increase in
the level of HIF-1?, the addition of 100 ?M AA still decreased the amount of HIF-1?
by nearly 90 %. There are at least two explanations for this unexpected result. One
is that AA has additional modes of action, other than affecting prolyl hydroxylase,
which result in a decrease in the HIF-1? protein. The other possibility is that in
the absence of PHD2, the isozyme PHD1 can be stimulated by AA and result in the targeting
of HIF-1? for degradation by the proteasome.

Regardless of the mechanism for the ability of AA and A2P to decrease HIF-1? levels
and inhibit HIF transcriptional activity, the important question is whether blocking
the HIF pathway decreases some of the malignant properties of the WM9 metastatic melanoma
cells. We addressed this question by measuring the ability of WM9 cells to invade
through Matrigel. A2P was able to inhibit invasion by 50 %. A2P did not inhibit the
proliferation of WM9 cells (data not shown). These findings, together with our previous
work 26] demonstrating that siRNA knockdown of HIF-1? also inhibits invasion through Matrigel,
suggests that AA affects the invasive ability of these metastatic cells through a
decrease in HIF-1?/HIF activity.

Although our studies used established human melanoma cell lines, there are some in
vivo studies that link AA levels to tumor aggressiveness. Low AA levels are associated
with increased HIF-1? levels and HIF stimulated gene products in human endometrial
tumors 29]. In contrast, increased tumor AA is associated with longer disease-free survival
and decreased HIF-1? and HIF stimulated gene products in human colorectal tumors.
Specifically in melanoma there is decreased plasma ascorbate levels in stage IV melanoma
patients 31] while an epidemiological study found an association between dietary vitamin C (AA)
and the risk of cutaneous melanoma in a Northern Italian population 52]. IL-2 treatment of melanoma is unfortunately associated with severe toxicity and
it causes a large decrease in circulating levels of AA. A clinical trial has been
proposed to assess the use of intravenous AA as an adjuvant to IL-2 treatment of melanoma
53]. Thus AA has many potential roles and uses in human melanoma. The next step will
be pre-clinical investigations of AA/A2P and HIF-1?/HIF activity in animal models
that most closely recapitulate the initiation and progression of human melanoma.