Well, it turns out the situation is more complex. I had assumed the answer was what rwst suggests or something to do with osmotic pressure. It seems that we don't really know that well.
In a paper from 1991, Chi and Wu suggest the following possible mechanisms :
Membrane fusion during the shedding of exovesicles might produce a transient decrease of the permeability barrier.
Increase of lipid dynamics by the alcohol could decrease the packing of the
bilayer. The membrane barrier behaves like a soft polymer, which can sieve solutes. The meshes in the polymer might become larger if its packing density is
reduced.
Lateral phase separation of lipids could induce packing defects in the lipid domain. This has been observed for long chain alcohols and postulated to be responsible for the increase of membrane permeability by amphiphiles.
Increase of the dielectric constant of the membrane by the alcohol would also increase the partition of hydrophilic solutes into the membranes. Such an increase has been postulated to be responsible for the increase of the permeability by aliphatic alcohols.
Modification of the intrinsic membrane domain might follow modification of the membrane skeleton by the alcohol. Accordingly, aggregation of intrinsic proteins might cause membrane modification mentioned under point 2 to 4.
The authors state that it is not possible to decide between the
various possibilities, but they seem to prefer point 5:
Although it is not possible to decide between the various
possibilities from the present data, we showed that the
release of membrane fragments from ethanol-treated RBC was not a
requirement for the creation of membrane pores since it occurred at a
time much later than the detection of K ÷ leakage. In addition, we
found that changes of membane rheological properties preceded the
permeability increase. These properties have been related to
the membrane skeletal protein spectrin. Moreover, ethanol has
been shown to affect the skeleton. The processes leading to
the formation of pores in ethanol-treated RBC may thus relate to a
deranged cytoskeletal network, followed by the aforementioned
alteration of membrane properties.
The plot thickens, apparently, low concentrations of alcohol protect erythrocytes from hemolysis while higher concentrations can cause it. The following are extracts from Tyulina et al:
The seeming paradox between the direct haemolytic effect of ethanol on
erythrocytes (Fig. 1) and the stabilizing effect of ethanol on
erythrocytes undergoing NaOCl-induced haemolysis (Fig. 2) could be
explained by the relatively small destabilizing effect of ethanol
which is observed (<1% haemolysis) over 16 h. This effect would be
negligible in the short time period (generally <10 min) assay for
NaOCl-induced haemolysis where 100% of the cells are haemolysed. An
alternative explanation for this paradox is that the mechanisms of
haemolysis induced by ethanol and NaOCl are different.
It therefore appears that ethanol does not induce significant
oxidative stress in the human erythrocyte, and these data are in
agreement with previous studies (Seeman et al., 1971), in which it was
found that low ethanol concentrations could protect erythrocytes
against haemolysis. Although the mechanism for this protective effect
is unknown, it has previously been suggested (Halliwell and
Gutteridge, 1999) that ethanol can serve as a hydrogen donor in the
elimination of the hydroxyl radical with formation of water and the
2-hydroxyethyl radical.
The authors then state (emphasis mine):
In summary, we conclude that the damage to erythrocytes which occurs
on in vitro exposure to ethanol may be caused, at least in part, by
unmetabolized ethanol directly, rather than by the oxidation of
ethanol to acetaldehyde or its conversion to FAEE.
I would guess this "direct effect" is something very much like what rwst suggested but the fact that the authors, who clearly work in this field, do not say so makes me think that the situation is more complex.
So, in conclusion, the exact details of alcohol's hemolytic properties don't seem to be understood in great detail. Admittedly, neither of these articles is very recent, if anyone can find a more up to date account I would love to read it.
REFERENCES:
Chi LM, Wu WG. Mechanism of hemolysis of red blood cell mediated by ethanol.
Biochim Biophys Acta. 1991 Feb 11;1062(1):46-50.
Tyulina OV, Prokopieva VD, Dodd RD, Hawkins JR, Clay SW, Wilson DO, Boldyrev
AA, Johnson P. In vitro effects of ethanol, acetaldehyde and fatty acid ethyl
esters on human erythrocytes. Alcohol Alcohol. 2002 Mar-Apr;37(2):179-86.
Trandum C, Westh P, Jørgensen K, Mouritsen OG. Association of ethanol with
lipid membranes containing cholesterol, sphingomyelin and ganglioside: a
titration calorimetry study. Biochim Biophys Acta. 1999 Aug 20;1420(1-2):179-88.
