The underlying assumption in your question - that the frequency being measured is the rate at which electrons reverse polarity - is incorrect. The frequency of a signal at the transmitter, receiver, or anywhere in between physically corresponds to the cyclic arrival of a voltage.
For example, in a digital application using amplitude modulation (let's assume on-off keying for simplicity), you could measure frequency by the number of 'on' pulses you detect per unit time. In RF communications, this might correspond to a logic-high voltage, or in optical communications it might correspond to the arrival of a large number of photons. In the ideal case, a logic-low or off state would correspond to a voltage of zero or the arrival of no photons, but dark currents and the imperfections of modulators rarely make that the case.
In terms of implementation, a straightforward and simple implementation to the transmission of two separate RF frequencies on a single medium (a copper wire) is by the use of two complete transmitter chains to encode the data at the two distinct carrier frequencies, and then the use of an RF combiner to get the two outputs from the transmitters onto a single copper wire. The receiver can be implemented a number of ways, but a simplistic method would be to use an RF power divider to create two copies of the signal, and then use a high pass filter on one and a low pass filter on the other. You can then continue with the normal receiver chain.
As others have said, multiple frequencies can be present on a wire at the same time. The instantaneous presence of multiple frequencies does not indicate multiple voltages though; there will necessarily be a single voltage at any given point on the wire (so long as the voltage is defined between that point and a common reference, typically ground). Over a span of time though, you can construct a signal by sampling at regular intervals. That signal will not look like a normal sinusoidal wave if multiple frequencies are present though, due to the principle of superposition. If you pick two carrier frequencies, let's say 5 kHz and 5 MHz, modulate data onto both and then sum the resultant modulated signals, you might be presented with a very peculiar signal in the time domain. If you apply a Fourier Transform though and look at the signal in the frequency domain, you would see a strong signal at 5 kHz, a strong signal at 5 MHz, and then a smattering of other frequencies around the carrier frequencies to account for the modulated data.