To our knowledge, this is the first demonstration that PFAAs do indeed have direct, developmental neurotoxicant actions and that they target specific events in neural cell differentiation. In general, the rank order of adverse effects was PFOSA > PFOS > PFBS ≈ PFOA. However, superimposed on this scheme, the various agents differed in their underlying mechanisms and specific outcomes, indicating that all PFAAs are not the same in their impact on neurodevelopment and that it is unlikely that there is one simple, shared mechanism by which they all produce their effects.
The greater toxicity of PFOSA can be partially attributed to its less hydrophilic nature. Because the other agents are free acids (PFOA) or sulfonic acids (PFOS, PFBS), PFOSA will more readily cross the cell membrane, achieving higher intracellular levels. By itself, this finding gives important information readily translatable to developmental neurotoxicity in vivo
: PFAAs that are more hydrophobic or that form less polar metabolites are likely to be more problematic, especially since the same physicochemical properties govern passage across the placental and blood–brain barriers. Nevertheless, pharmacokinetic differences cannot account for the disparities in actions among the various PFAAs: PFOS elicited larger changes than PFBS, despite the fact that both are sulfonic acids, and all four agents had differing, even opposite, actions on neurotransmitter phenotypes. Certainly, one likely possibility for the greater toxicity of PFOSA is its ability to uncouple mitochondrial oxidative function, whereas PFOS and PFOA act simply as surfactants at the mitochondrial membrane (Starkov and Wallace 2002
). This feature may contribute to disparities in the targeting of specific events in cell differentiation that are affected by oxidative stress and other downstream events linked to mitochondrial dysfunction.
PFOSA was the only one of the agents tested that matched or exceeded the ability of CPF to inhibit DNA synthesis in undifferentiated cells. Likewise, PFOSA elicited the greatest degree of oxidative stress and cell loss, regardless of whether cells were undifferentiated or differentiating. Nevertheless, even with PFOSA we did not see significant loss of cell viability until the concentration was raised to 250 μM, implying that there are factors other than cytotoxicity that contribute to the net effects. In fact, our results point to strong promotion of the switch from cell replication to cell differentiation, which would also contribute to a reduction in DNA synthesis and in cell numbers. The protein ratios provide support for this interpretation. First, the protein/DNA ratio increased with PFOSA treatment at concentrations below the threshold for cytotoxicity, indicating an increase in cell size rather than the suppression of growth that would be expected from cytotoxic actions. Second, the membrane/total protein ratio also showed a rise with subtoxic PFOSA treatments, indicating augmented membrane complexity, which is commensurate with generation of neurites and cellular organelles that accompanies differentiation. The final proof of a prodifferentiation effect can be seen from the strong promotion of both neurotransmitter phenotypes, evidenced by marked increases in both TH and ChAT. The differentiation pattern triggered by PFOSA is not, however, a normal one: At low concentrations, the TH/ChAT ratio was slightly, but significantly decreased, whereas at high concentrations, it rose markedly. This means that PFOSA alters the differentiation fate of the cells, switching them weakly to the ACh phenotype at low concentrations, and strongly to the DA phenotype at high concentrations. If similar effects happen in vivo
, it might be expected that neurons will differentiate into inappropriate phenotypes; this would lead to miswiring of neural circuits, with presynaptic projections for a given neurotransmitter juxtaposed to post-synaptic elements containing incorrect receptors, resulting in nonfunctional synapses. Given the biphasic dose–response curve, the outcomes would be very different for an individual exposed to low doses that promote the ACh phenotype as opposed to an individual receiving higher exposures that promote the DA phenotype. Such switching has already been seen with other agents that produce phenotype shifts in the PC12 model, notably CPF (Barone et al. 2000
; Dam et al. 1999
; Jameson et al. 2006b
; Pope 1999
; Rice and Barone 2000
; Slotkin 2004a
; Vidair 2004
). The biphasic nature of the concentration–effect relationship suggests that there are at least two separate mechanisms operating to turn on expression of neurotransmitter phenotypes. Oxidative stress, which was prominent for PFOSA, is likely to be one contributory factor, since oxidative stress itself is a prodifferentiation signal (Katoh et al. 1997
). However, it is also possible that PFOSA switches the phenotypes by turning specific DA- or ACh-related genes on or off inappropriately, as has been shown for the organophosphates (Slotkin and Seidler 2007
Although PFOS also inhibited DNA synthesis in undifferentiated PC12 cells, its effects were less than those seen with the positive test compound (CPF) and far smaller than those seen with PFOSA. There was no parallel reduction in DNA content, which is not surprising given the small effect on DNA synthesis and the fact that a 24-hr exposure is less than a doubling time for PC12 cell replication. Notably, we did not find any reductions in cell numbers even with a 6-day exposure of differentiating cells, showing that the effects of PFOS are indeed fundamentally different from those of PFOSA. This was further confirmed by a lack of any evidence for a global prodifferentiation effect because there were no changes in the protein ratios and only a small degree of oxidative stress. Nonetheless, PFOS altered the phenotypic fate of the cells, promoting the ACh phenotype at the expense of the DA phenotype; the effect was again biphasic, as the increase in ChAT regressed back to normal as the PFOS concentration was raised above the point where lipid peroxidation and cytotoxicity emerged. There are two important points made by these findings: a) The impact of PFOS on neurotransmitter phenotype is radically different from that of PFOSA; and b) some of the effects on differentiation are distinct from those related to oxidative stress and cytotoxicity, and occupy a part of the concentration–effect curve below the thresholds for those common adverse events.
Of the four agents, PFOA and PFBS had the least effect on most of the parameters connoting cell acquisition and growth. Neither agent produced any substantial inhibition of DNA synthesis or cell loss in undifferentiated cells, and a small decrease in DNA in differentiating cells was found only at the highest concentration, and only for PFOA. Although both agents produced small but significant decreases in viability at concentrations of 100 or 250 μM, the effect was obviously insufficient to have any impact on the total number of cells remaining in the culture. Similarly, PFOA had no effect on protein ratios, and PFBS had only a small effect on total protein/DNA without any impact on membrane/total protein. For phenotypic outcomes, the two agents were substantially different from PFOSA and PFOS, both in the type and magnitude of effects. PFOA caused a slight reduction in TH and, consequently, a minor shift favoring the ACh phenotype (reduced TH/ChAT ratio). In contrast, PFBS retarded differentiation into both phenotypes, an effect not seen with any other agent; accordingly, although the TH/ChAT ratio was unaffected by PFBS, both TH and ChAT were significantly reduced, indicating a likelihood of impaired function for both neurotransmitters. Once again, this demonstrates that the effect on phenotypic fate of the cells is distinct from any other effects on cell replication, growth, or viability.
Our results thus point to the potential for PFAAs to evoke developmental neurotoxicity through direct actions on replicating and differentiating neurons, effects distinct from the indirect consequences of endocrine disruption or metabolic or other secondary mechanisms. Importantly, the various PFAAs differ not only in their propensity to elicit outright neurotoxicity or oxidative stress but also in their ability to augment or suppress specific neurotransmitter phenotypes, thus shifting the differentiation fate of the neuron. These effects can even be in opposite directions for the various PFAAs, and because the concentration–effect curve is biphasic, different levels of exposure can be expected to produce disparate outcomes directed toward a given neurotransmitter phenotype. The fact that there are stark differences in developmental neurotoxicant actions among otherwise similar PFAAs means that it is important to take developmental neurotoxicity into account in the design of future members of this class. As demonstrated here, this can be aided by the use of in vitro models that permit rapid and cost-effective screening for developmental neurotoxicity.