as operators that execute transformations
of the probabilities. Moving the mathemat-
nity toward an ability to model and predict
this matrix may be an exciting ‘‘next fron-
tier’’ waiting beyond the present important
Afkarian, M., Sedy, J.R., Yang, J., Jacobson, N.G.,
Cereb, N., Yang, S.Y., Murphy, T.L., and Murphy,
K.M. (2002). Nat. Immunol. 3, 549–557.
Mullen, A.C., High, F.A., Hutchins, A.S., Lee, H.W.,
Villarino, A.V., Livingston, D.M., Kung, A.L., Cereb,
N., Yao, T.P., Yang, S.Y., and Reiner, S.L. (2001).
Science 292, 1907–1910.
Rothenberg, E.V. (2007). Nat. Immunol. 8, 441–
Schulz, E.G., Mariani, L., Radbruch, A., and
Szabo, S.J., Kim, S.T., Costa, G.L., Zhang, X.,
Fathman, C.G., and Glimcher, L.H. (2000). Cell
Szabo, S.J., Sullivan, B.M., Peng, S.L., and
Glimcher, L.H. (2003). Annu. Rev. Immunol. 21,
Tomlin, C.J., and Axelrod, J.D. (2007). Nat. Rev.
Genet. 8, 331–340.
Turing, A.M. (1952). Philos. Trans. R. Soc. Lond. B
Biol. Sci. 237, 37–72.
(2003). Immunity 18, 415–428.
Usui, T., Preiss, J.C., Kanno, Y., Yao, Z.J., Bream,
J.H., O’Shea, J.J., and Strober, W. (2006). J. Exp.
Med. 203, 755–766.
How to Polymerize in Order to Survive
Eckhard R. Podack1,*
1Department of Microbiology and Immunology, University of Miami, Leonard Miller School of Medicine, Miami, FL 33136, USA
Perforation of membranes and pore formation is mediated by polymerization of proteins of the immune
system, complement C9 and Perforin, which share the conserved MACPF domain. In this issue of Immunity,
Baran et al. (2009) identify the molecular mechanism initiating polymerization as charge interactions in the
Survival of the fittest, Darwin’s sound bite
sary year of his birth, certainly applies
to the biological warfare that we—our
immune systems—engage in, day after
day, lifelong. The immune system uses
sophisticated weapons to recognize and
combat invading pathogens, just as the
invaders continue to improve their weap-
onry. In this ongoing arms race, chemical
attack and counterattack has long been
appreciated as common practice, taking
the form of toxins used by pathogens,
and reactive oxygen and various types
of hydrolases used by immune cells, to
kill invaders. Physical attack mechanisms
puncturing holes into membranes have
been recognized later as part of the
biological arsenal used for attack and
defense. Physical membrane attack is
mediated by specialized proteins that
can insert into and polymerize within the
lipid bilayer of membranes to form giant
transmembrane pores, which are, for a
cell, more akin to bomb craters. The
detailed molecular mechanism by which
gered, however, has remained hidden—
until now. In this issue of Immunity, Baran
et al. (2009) lift the veil of this long-kept
secret and describe how polymerization
occurs for Perforin, the membrane-attack
and pore-forming protein of cytotoxic T
lymphocytes (CTLs) and natural killer
In a painstaking mutagenic analysis,
Baran et al. (2009) identify three charged
amino acids located in the MACPF (mem-
brane attack complex of complement-
perforin) domain of Perforin as being
critical for triggering the polymerization
reaction leading to pore formation. (For
more about the MACPF domain, see
below.) Importantly, the charged residues
of monomeric perforin that are needed for
polymerization must be of opposite polar-
ization and on opposite sides of a mono-
mer. This arrangement allows a specific
positive residue on one monomer to
react with a specific negative residue on
the opposite side of a second monomer
(Figure 1) to initiate polymerization. Baran
et al. (2009) also show in the paper that
polymerization does not happen if the
left and right residues on one monomer
have the same charge. If, however, the
charges on each monomer are reversed,
and remain opposite in polarity and posi-
tion to each other, polymerization func-
tions again. Charge interaction thus is
the initiating reaction for polymerization,
followed by close alignment of two mono-
mers and refolding of part of the molecule
(orange), which mediates its membrane
insertion as a dimer, or subsequent higher
polymer. The polymerization reaction
continues within the membrane until it is
(top view) or rather a hollow cylinder (side
view), creating the bomb crater in the
membrane mentioned above. Because
there are many charged residues in the
MACPF domain of perforin, it is worth
pointing out that only those that are in
the correct position and orientation as
the critical polymerization reaction.
It is self-evident that the formation
of transmembrane pores requires large
amounts of energy for the displacement
of lipid in the membrane and for creation
Immunity 30, May 22, 2009 ª2009 Elsevier Inc.
tion proceeds without energy input from
or covalent changes. It has long been
a known fact that purified perforin, in the
presence of Ca ions and phospholipid
merize within the lipid membrane, thereby
forming large water-filled pores. The only
source of energy thus can be derived
from the molecular interaction of two
monomers. This interaction apparently
provides sufficient energy for refolding
of the monomers and for driving mem-
brane insertion and lipid displacement.
Lipid displacement in turn is caused by
the amphipathic nature of the refolded,
membrane-inserted sequences that are
believed to form amphipathic b-pleated
sheets (orange in Figure 1), which again
are polarized, but now in the sense that
one side (the outside) is hydrophobic
and the other (the inside) is hydrophilic.
tion of at least two monomers forming
a dimer and inserting into the membrane.
The smallest channel that can be formed
therefore must be at least a dimer. Sup-
porting the model of polymerization for
pore formation by complement compo-
nent C9 and Perforin, different size chan-
nels have been documented for both
pore formers by measuring conductance
increases, consistent with the increasing
size of the growing polymers (Young
et al. (2009) also is in full agreement with
As always in science, new discoveries
are built on top of previous findings. The
polymerization reaction for transmem-
brane pore formation was initially des-
cribed for complement component C9 in
1982 (Podack and Tschopp, 1982) and
shortly thereafter followed by the descrip-
tion of Perforin polymerization (Blumen-
thal et al., 1984; Podack and Dennert,
1983). It took more than 10 years until
physical membrane attack by CTLs and
NK cells via Perforin was taken seriously
by most immunologists and appreciated
as an important defense mechanism.
Only when the first Perforin gene deletion
was achieved and the Perforin-deficient
mouse shown to be defenseless against
certain virus infections was the impor-
tance of Perforin fully appreciated (Kagi
et al., 1994). Implications for a role of
Perforin in immune homeostasis were
reported in mice (Spielman et al., 1998)
patients who suffer from lethal familial
lymphohistiocytosis, apparently caused
by the inability of Perforin-deficient CTLs
to eliminate antigen presenting cells after
viral infection (Stepp et al., 1999).
C9 and Perforin are the founders of
the family of MACPF-domain-containing
proteins, a family that continues to grow
in our genomic era, with sequences from
virtually all species including prokaryotes
and plants. The MACPF domain has
helix sequence, previously thought to
enter the membrane, is involved in the
newer data. These insights have become
possible with the recent crystallization
of the MACPF domain of C8a of comple-
ment and delineation of its 3D structure
(Hadders et al., 2007; Slade et al., 2008).
The crystal structure of the MACPF
domain of C8a also revealed its structural
homology with a family of bacterial toxins
that form pores in cholesterol-containing
membranes—another example of the
arms race in which the invention of a new
weapon is soon copied (stolen?) by the
The recent crystal structure of the
MACPF domain and the analysis of the
mechanism of polymerization in Baran
et al.’s paper allow us to ask and answer
interesting new questions, address new
possibilities, and manipulate the polymer-
ization process for our benefit. Is the
ionic-polarized interaction of the MACPF
domain required for polymerization of all
members of the family? What are the
structural constraints for the precise loca-
tion of the polarized residues within the
MACPF domain? Is the MACPF domain
always a pore-forming domain, or is poly-
merization used also for other purposes?
Not least on the list of directions for future
study: with the structural knowledge of
how polymerization is initiated in molec-
ular and atomic terms, it will be possible
to develop targeted drugs that inhibit
polymerization of Perforin. This may be
a boon for blocking exuberant killer cells
when they cause too much collateral
damage or when they are on the loose in
Figure 1. Polymerization and Refolding of Perforin, Driven by Charge Interactions
Top row, simplified schematic view of the results in this issue (Baran et al., 2009); charge interactions on
nonmutated Perforin (red and blue) initiate contact of Perforin monomers resulting in refolding and inser-
tion of hydrophobic domains (orange) into the lipid bilayer. Mutation of Perforin to bear identical charges
blocks polymerization (middle row), whereas reversing charges permits polymerization again (third row).
The refolded part of Perforin (orange in the schematic above) forms the transmembrane part of the pore
(schematically represented at the bottom of the figure), most likely as amphiphilic beta-barrel.
Immunity 30, May 22, 2009 ª2009 Elsevier Inc.
autoimmune or autoaggressive disease.
The new climax of research on Perforin,
as reported here, may soon be followed
by the development of novel therapeutics
targeted to enhance or diminish Perforin
cytotoxicity, to enhance our fitness and
Baran, K., Dunstone, M., Jenny, D., Ciccone, A.,
Brownie, K.A., Clarke, C.J.P., Lukoyanova, N.,
Saibil, H., Whisstock, J.C., Voskoboinik, I., and
Trapani, J.A. (2009). Immunity 30, this issue,
Blumenthal, R., Millard, P.J., Henkart, M.P., Rey-
nolds, C.W., and Henkart, P.A. (1984). Proc. Natl.
Acad. Sci. USA 81, 5551–5555.
Hadders, M.A., Beringer, D.X., and Gros, P. (2007).
Science 317, 1552–1554.
Kagi, D., Ledermann, B., Burki, K., Seiler, P., Oder-
matt, B., Olsen, K.J., Podack, E.R., Zinkernagel,
R.M., and Hengartner, H. (1994). Nature 369,
Podack, E.R., and Dennert, G. (1983). Nature 302,
Podack, E.R., and Tschopp, J. (1982). Proc. Natl.
Acad. Sci. USA 79, 574–578.
Slade, D.J., Lovelace, L.L., Chruszcz, M., Minor,
W., Lebioda, L., and Sodetz, J.M. (2008). J. Mol.
Biol. 379, 331–342.
Spielman, J., Lee, R.K., and Podack, E.R. (1998).
J. Immunol. 161, 7063–7070.
Stepp, S.E., Dufourcq-Lagelouse, R., Le Deist, F.,
Bhawan, S., Certain, S., Mathew, P.A., Henter,
J.I., Bennett, M., Fischer, A., de Saint Basile, G.,
and Kumar, V. (1999). Science 286, 1957–1959.
Young, J.D., Cohn, Z.A., and Podack, E.R. (1986).
Science 233, 184–190.
Lymphoid Organs for Peritoneal Cavity
Immune Response: Milky Spots
Reina E. Mebius1,*
1Department of Molecular Cell Biology and Immunology, VUMC, van der Boechorststraat 7, 1081 BT Amsterdam, The Netherlands
Milky spots are located in the omentum of the peritoneal cavity and their classification as lymphoid organs
has been debated. In this issue of Immunity, Rangel-Moreno et al. (2009) provide compelling data to consider
them as unique secondary lymphoid organs.
The omentum is formed by a double layer
of mesothelial cells that connects the
liams and White, 1986). Embedded within
are clusters of leukocytes, called milky
spots. Milky spots are mainly composed
of macrophages and B1 cells, resembling
B cells that can be distinguished from
conventional B (B2) cells by expression of
distinct cell-surface markers and antigen
receptors that can bind common bacterial
epitopes, as well as by their recognized
potential to produce natural antibodies
that provide a first protection to bacterial
infections. B1 cells are localized in distinct
anatomical locations, such as the perito-
neal and pleural cavities, but are also
present in the spleen. For their localization
to these body cavities, but not the spleen,
the chemokine CXCL13 is of great impor-
tance because in Cxcl13?/?mice, B1 cells
as well as milky spots are absent from the
pleural and peritoneal cavities (Ansel
etal., 2002). As a consequence, Cxcl13?/?
mice, as well as mice deficient for the
CXCL13 receptor, CXCR5, have strongly
reduced titers of natural antibodies (Ansel
milky spots are an important source of
Because milky spots mainly consist
of macrophages and B1 cells, and are
described to lack dendritic cells as well as
follicular dendritic cells, controversy exists
asto whether tomarkthese milky spots as
Immunity, Rangel-Moreno et al. (2009)
addressed the immunological potential of
spleen, and Peyer’s patches. Here, they
used splenectomized lymphotoxin-alpha
(LTa)-deficient mice (Lta?/?), which as
a result of their deficiency already lacked
lymph nodes and Peyer’s patches. These
animals, devoid of secondary lymphoid
organs, were reconstituted with wild-type
bone marrow (SLP mice) and compared
to irradiated C57BL/6 mice that were simi-
larly reconstituted with wild-type bone
marrow. Antigens injected into the perito-
neal cavity of SLP mice were shown to
collect in the milky spots, resulting in the
generation of antigen-specific antibodies.
Furthermore, germinal center B cell res-
ponses, supporting isotype switching,
somatic hypermutation, and some affinity
in response to intraperitoneally injected
antigens, could be observed in the milky
pied by locally activated lymphocytes, but
also lymphocytes that encounter antigens
elsewhere were shown to recirculate
through the milky spots. These experi-
ments clearly identified the milky spots
as part of the general surveillance route
of antigen-experienced lymphocytes in
search for their antigen.
What does it take to enter and leave the
milky spots? Milky spots contain high
endothelial venules (HEVs) that express
the peripheral lymph node addressin
(PNAd) as well as the mucosal addressin
(MAdCAM-1), permitting entry of lympho-
cytes from the bloodstream into the milky
Immunity 30, May 22, 2009 ª2009 Elsevier Inc.