Five Steps of Septin Morphogenesis During Zygote Formation
When haploid yeast of opposite mating type are mixed at room temperature, zygotes form after 1.5–2.0 hr. To establish the relative timing of cis-to-trans redistribution of parental proteins and organelles, we used time-lapse microscopy and visualized fluorescent marker proteins. Population-based estimates of relative timing agree with time-lapse observations, but the temporal dispersion of these events makes it more informative to use time-lapse, which also can illustrate the suddenness of redistribution. The selected images and time-lapse series illustrated below are in all cases representative of examination of at least twenty cells.
We initially observed that redistribution of distinct organelles and supramolecular complexes is by no means synchronous. We therefore have inquired whether cytoskeletal barriers partition the cytoplasm, beginning with septins.
After treatment of MAT a haploid cells with mating factor for 2–3 hr, the tagged septin, GFP-Cdc3p, forms a collar at the cell cortex distal to the tip of the mating projection, as previously described (Ford and Pringle, 1991
; Kim et al., 1991
; Longtine et al., 1998
) (). This collar has a composite organization in which lobes are joined at their apical ends and become increasingly splayed as they extend distally. A pool of diffuse cytoplasmic fluorescence is also evident.
When a mating pair meets, the two plasma membranes form a flattened “zone of contact” (Byers and Goetsch, 1975
). In crosses in which one partner expresses GFP-Cdc3p, contact is followed by entry of a diffuse signal into the acceptor cell. Within 5 minutes, the collar in the acceptor cell then becomes symmetrically labeled (, 4–6 min time points). Such behavior is expected if a soluble pool of GFP-Cdc3p permeates cis
, and if assembly of the collar is dynamic.
Just prior to nuclear contact, a GFP-Cdc3p-positive structure appears at the interface between the two parental domains (e.g. ). It can also be detected in cells expressing other tagged septins (GFP-Cdc10p, GFP-Cdc11p or GFP-Cdc12p) (not shown). In face view, it appears as an annulus ( – insert). Its diameter is somewhat greater than the septin hourglass at the bud neck, which is brighter and wider. In typical experiments in which cells expressing GFP-Cdc3p are examined 2 hr after mixing, this structure is evident in >90% of zygotes for which plasma membrane fusion has occurred. Parental nuclei congress and fuse with each other when it is already in place (). The concentration of GFP-Cdc3 at the midzone is obvious for ~15 min. A patch of GFP-Cdc3p at the site of future bud emergence then appears ().
GFP-Myo1p is not detected at the cortex of the mating projection; however, it also forms an annulus in the midzone (). Judging from the distribution of GFP-tagged Act1p, and the actin-binding protein, Abp140/Trm140, and from staining fixed preparations with rhodamine-phalloidin, actin patches and cables are widespread but are not characteristically concentrated or oriented in the midzone at this time (see below).
Photobleaching (FRAP) of GFP-Cdc3 in zygotes shows that the cortical collar, annulus, patch, and bud neck filaments are all dynamic. Since the estimates of half-time for recovery (13.5+/−3.4 sec – 27.1+/−8.7 sec, n=5–21) and mobile fraction (28+/−2.8% – 38+/−8.5%, n=5–21, S.D.) are relatively uniform among these structures, the successive redistribution of septins is likely to result from the progressive elimination and/or appearance of binding sites. Additional evidence of the dynamic nature of the annulus comes from FLIP experiments: repeated photobleaching of the diffuse pool of GFP-Cdc3p in the zygote cytoplasm (avoiding the annulus itself) progressively weakens the signal in the annulus (Figure S1
We therefore suggest the following sequence of septin morphogenetic intermediates (): 0: the collar of the mating projection, 1: the symmetric collar of early zygotes, 2: the annulus, 3: the patch at the site of bud formation, 4: the bud neck itself. As is explained below, these events are concurrent with changes in the actin cytoskeleton.
Since the medial concentration of septins becomes most obvious when parental nuclei establish contact, we explored the possibility that the annulus has a continuing association with the nucleus. In support of this possibility, we observe that 1) a narrowed segment of the nuclear envelope spans the midpoint of the zygote even after karyogamy and displacement of the chromatin mass to one side (), 2) nuclear pores are generally not detected at this point (), and 3) the annulus encircles the narrowed segment ().
Flux of Polysomes across the Midzone is Delayed Relative to Soluble Tracers
Shortly after cell-cell contact is established, soluble DsRed (~120kDa) – like GFP-Cdc3 - suddenly redistributes from the donor to acceptor cell. Surprisingly, at this point polysomes (including Rps3p-GFP or Rpl25p-mCherry) remain in their initial parental domain (). In fact, they begin to redistribute only when nuclei are about to establish contact, well after rupture of the plasma membrane, as judged by the apical elimination of the plasma membrane protein, Mid2p-GFP () that is present at the zone of contact prior to rupture. The equilibration of polysomes then requires ~10 min. The slow pace of these events suggests the presence of a barrier, given the apparent full disruption of the plasma membrane, the significant mobility of polysomes (t1/2
for recovery after photobleaching ~16 sec (Figure S2
)), and electron micrographs that show a 0.5–1 micron gap between the nuclear envelope and the cortex of the zygote, e.g. (Byers and Goetsch, 1975
As illustrated above, while fluorescent polysomes gradually shift from cis
, there is a sharp discontinuity in their signal intensity. To learn whether the nucleus is responsible for this discontinuity and impedes transit through the midzone, we studied kar1
× wt crosses, in which nuclei do not congress, e.g. (Molk and Bloom, 2006
) (). The presence of the nucleus at the midzone does not appear to contribute, since a) a sharp discontinuity in polysome distribution is evident in such crosses (Figure S3
), and b) there is no significant acceleration of the rate of polysome redistribution in kar1
× wt crosses by comparison to wt × wt crosses. Moreover, the redistribution of polysomes still precedes redistribution of the mitochondrial signal – as for wt × wt crosses (Figure S4
Impact of septin mutants and nuclear congression on polysome flux
Since septins concentrate at the zygote midzone, we also asked whether they contribute to the slow pace of flux. Indeed, temperature increase from 23°C to 37°C causes a modest increase in the rate of flux in crosses between temperature-sensitive (ts) conditional septin mutants (cdc12-6), while no comparable increase is seen when wt control crosses are studied at the same temperatures ().
Flux of the Prion, Sup35 [PSI+], is Delayed Relative to Sup35 [psi−]
To learn whether Sup35p [PSI+
] readily traverses the midzone, we followed Sup35p-GFP in crosses in which one mating partner expresses a functional integrated copy of Sup35p-GFP from the MFA1
promoter, that is turned off upon cell fusion. The same partner also expresses soluble DsRed or Rpl25-mCherry. In [psi−
] cells, cytoplasmic Sup35p-GFP is smoothly distributed and can diffuse freely, while in [PSI+
] cells the signal is generally inhomogeneous (Greene et al., 2009
; Kawai-Noma et al., 2006
; Satpute-Krishnan and Serio, 2005
). In the “strong” [PSI
+] strains that we use, Sup35-GFP has a mottled/irregular appearance at steady-state. The designation “strong” signifies that translation termination defects are suppressed more efficiently than by “weak” forms.
Redistribution of aggregated Sup35p-GFP in crosses between [PSI+
] cells is restricted at the midzone, judging from examination of intermediate time points (). This is reminiscent of previous studies of the [HET-s] prion (Mathur et al., 2012
). The restriction is not simply due to the position of the nucleus, since an equivalent discontinuity is also evident in wt × kar1
crosses in which the lack of karyogamy is ensured by following a marker of the ER/nuclear envelope (mRFP-HDEL) (). Moreover, in such crosses, redistribution of Sup35 still occurs before redistribution of mitochondria (not shown).
Further experiments help characterize the medial barrier and the in vivo hydrodynamic properties of Sup35p. First, Sup35p-GFP (91kDa) [psi−] and soluble DsRed (~120kDa) have similar hydrodynamic properties in vivo: both transfer over the same period of time in crosses between [psi−] cells (). Transfer of Sup35p-GFP is essentially complete before initiation of polysome equilibration ().
Moreover, progressive coalescence of diffuse Sup35p can be detected, possibly as a reflection of conformational maturation. Thus, when Sup35p-GFP from a [psi−
] “donor” enters a [PSI+
] “acceptor” environment, tiny just-discernable foci appear in the acceptor cell domain during the initial 30 minutes. These foci provide a first suggestion - at this resolution - of conformational change and possible oligomer/aggregate formation. The donor cell domain, as in studies of heteroallelic conversion of the HET-s and HET-S proteins (Mathur et al., 2012
), could receive non-fluorescent seeds of [PSI
+] Sup35 in such experiments. We do not, however, see progressive changes in the donor compartment over the same period of time, perhaps because the size of any seeds precludes their rapid exchange (). More visible foci and extensive “mottling” do appear in both domains, but this occurs only gradually over 1–2 hours, e.g. (). Curiously, an earlier study has described more rapid Sup35-GFP conversion during zygote formation (Satpute-Krishnan and Serio, 2005
Furthermore, the overall polydispersity of Sup35p in a [PSI+] environment appears comparable to that of polysomes, i.e. in [PSI+] × [PSI+] crosses in which both Sup35p-GFP and tagged polysomes are expressed by one parent, the transfer of Sup35p-GFP aggregates extends over at least as long a period as for polysomes ().
Mitochondrial Encounters and Fusion
Progressive Sequestration of Septins to the Bud Neck Parallels Mitochondrial Encounters
After polysome and prion flux and completion of nuclear fusion, tagged mitochondria extend precisely up to the midzone, as though abutting on an invisible barrier (). During this period their position is however not further restricted - as in haploid cells, they can move extensively. Only after a 15–30 min delay do matrix markers contributed by one parent quickly access much of the mitochondrial labyrinth of the trans
domain of the zygote (), presumably as a result of sudden fusion between the parental mitochondria (Hermann et al., 1998
; Nunnari et al., 1997
; Okamoto et al., 1998
) – see also Figure S5
. In , note that, prior to redistribution of the mitochondrial marker, a patch of GFP-Cdc3 appears at the site of bud formation (asterisk) – as in . The consistency of this order is evident in experiments in which a parent that expresses GFP-Cdc3p was crossed with a parent that expresses the matrix marker, Cox4-DsRed: At a time point when redistribution of the matrix marker had occurred in half of the zygotes, a GFP-Cdc3-positive cortical patch or bud neck was evident in all zygotes (83/83).
The timing of redistribution of DsRed-Cox4 and the sequential morphogenesis of septin-containing structures suggests a “sequestration” hypothesis: that the zygote midzone initially impedes encounter of parental mitochondria, that the site of incipient budding then recruits components from the midzone, including septins, and that the integrity of the midzone becomes so impaired that cis-trans encounter of parental mitochondria can occur. Relocation of selected proteins from the midzone to the bud neck could also cause secondary changes that promote redistribution – as is further discussed below.
Encounters of Mitochondria Require Actin Polymerization and Recruitment of Septins to the Bud Neck
To learn whether septin integrity affects encounter of parental mitochondria, we first evaluated redistribution of Cox4-DsRed in crosses between cdc12-6 strains. These “two-step crosses” were initiated at 23°C and then reincubated for up to 40 min at 37°C vs 23°C. As shown in , redistribution is little affected at 37°C vs 23°C for the cdc12-6 cross, and the rate is nearly identical at both temperatures for wt cells. In these experiments, multiple pools of septins are perturbed, i.e. any medial barrier could be weakened and any role for septins at the site of bud emergence could also be compromised.
We therefore studied redistribution of Cox4-DsRed in crosses of mutants that inhibit bud emergence (). In each case, one of the parents also expressed GFP-Cdc3p. Relevant mutants are a) an exocytosis ts mutant that stops budding, sec1-1
, e.g. (Togneri et al., 2006
), and b) the ts cyclin-dependent kinase mutant, cdc28-13
, that stops both budding and deposition of septins at the site of bud formation in haploid cells (Cid et al., 2001
). As expected, no zygotic buds formed in either cross at 37°C.
In sec1-1 × sec1-1 two-step crosses, a cortical patch of GFP-Cdc3p appeared in ~2/3 of zygotes within 40 min during the reincubation at 37°C and the annulus became less evident with time ( - left). Moreover, redistribution occurred at essentially the same rate at both 37°C and at 23°C (). In cdc28-13 × cdc28-13 crosses, the annulus is readily detected, but there was no cortical patch of GFP-Cdc3p after incubation at 37°C ( - right). Moreover, the cdc28-13 crosses consistently showed slower redistribution at 37°C than at 23°C ().
Parallel two-step crosses show that actin polymerization is required in order for redistribution of Cox4-DsRed: addition of latrunculin A during the second incubation halts redistribution (). This treatment does not cause an obvious change in the distribution of GFP-Cdc3 (not shown).
Impact of latrunculin A on redistribution of mitochondria
Thus, encounter and fusion of parental mitochondria are delayed when the septin annulus is conspicuous and bud neck filaments have not formed. In this sense, by checking on the progress of bud formation, the timing of encounters between parental mitochondria is adjusted.
The Nucleus Impedes Redistribution
To learn whether the presence of the nucleus at the midzone delays the encounter of parental mitochondria, we evaluated redistribution of Cox4-DsRed in kar1 × wt crosses, by comparison to wt × wt crosses, and observed that redistribution occurs earlier in the kar1 x wt crosses (). Both septins and the nucleus thus contribute to the delay of mitochondrial encounter and fusion.
Entry into Buds
Sup35p [PSI+] Enters All Buds
Some forms of Sup35p are not efficiently inherited during mitotic growth. Nevertheless, time-lapse observations of individual [PSI+
] × [PSI+
] and [PSI+
] × [psi−
] zygotes in which aggregated Sup35p-GFP is introduced from a [PSI+
] parent show that all buds – including the smallest that are encircled by septins - receive aggregated Sup35p-GFP (e.g. ), a process that is favored by fragmentation of prion units (Liebman and Chernoff, 2012
; Paushkin et al., 1996
). This is also true in [psi−
] × [PSI+
] crosses in which Sup35p-GFP is contributed by the [psi−
] parent. Indeed, there is no visible distinction between buds originating at the two distinct ends in [psi−
] × [PSI+
] crosses. Thus, although these aggregates of Sup35pstrong
-GFP are delayed at the midzone, and although there can be quantitative differences in the relative abundance of aggregates among cells, they are transmitted to all progeny. It will be of interest to investigate the extent to which the [RNQ
+] status of cells and other forms of Sup35 (e.g. weak vs
strong) may influence transit between parental domains and entry into buds.
Nucleoids Enter Nascent Terminal Buds and Mother-Bud Continuity of Mitochondria Continues Until Telophase
To learn whether delayed fusion of parental mitochondria causes terminal buds to be enriched in a single parental mitochondrial genome, we conducted crosses in which one parent expressed tagged proteins of mitochondrial nucleoids - Abf2p-GFP or Mgm101p-GFP (Kucej et al., 2008
; Meeusen et al., 1999
; Okamoto et al., 1998
) - and the other expressed Cox4-DsRed. Indeed, tagged nucleoid(s) consistently associate with adjacent nascent terminal buds well before cis-trans
fusion of mitochondria, e.g. ( and S6
The biased inheritance of mitochondrial genomes could signify that there is only a brief time window for association of mitochondria with terminal buds. Alternatively, mitochondria could retain continuity into the bud for an extended period of time, but nearby (cis
) nucleoids that enter early might outnumber nucleoids derived from the distant parent, or associate with a finite number of binding sites. It is therefore important to learn for how long mitochondria remain continuous across the bud neck. Published images show continuity when buds are present, e.g. (Boldogh et al., 2005
; Garcia-Rodriguez et al., 2009
; Weisman, 2006
Since buds form before entry of the nucleus, we have used two protocols to take this analysis a step further, showing that continuity of Cox4-GFP into buds continues when the nucleus spans the bud neck: 1) When the cell cycle is arrested by inactivating the mitotic exit network kinase, Dbf2p – (), and 2) After depletion of the activator of the anaphase promoting complex (APC), Cdc20p (Komarnitsky et al., 1998
) – (). In the latter case we have used FRAP to assess functional continuity. When the bud is bleached, the signal can a) diminish in the bud but not change in the zygote, b) immediately diminish in both bud and in the zygote, or c) recover in the bud only after a delay. Each outcome is observed with comparable frequency. Thus, although mother-bud continuity can be intermittent, it persists until after entry of the nucleus.
Mitochondria Enter Medial Buds Before Fusing
Both parental types of mitochondria are present in most diploid cells that originate from medial buds, although one type is lost (at random) within a few generations (Birky, 1978
; Dujon et al., 1974
; Okamoto et al., 1998
; Thomas and Wilkie, 1968
). As there has been no indication of whether parental mitochondria fuse with each other before entry into buds, we have studied early stages of medial bud emergence using parents that express Cox4-DsRed vs
Cox4-GFP. One readily finds examples in which both types of mitochondria extend to the bud neck but have not fused, suggesting that the two types of parental mitochondria generally fuse with each when they enter the bud – . Thus, as for non-medial events, fusion occurs when bud formation is already underway. Interestingly, fusion of parental vacuoles also does not occur before entry into medial buds (Weisman, 2006
The Septin Ring at the Bud Neck is Required for Actin Polarization
Why do parental mitochondria not fuse with each other long before bud formation ? Is a medial barrier strongly restrictive or do bud formation, the arrival of septins at the neck, and actin polarization also have a positive effect ? To explore this issue, we first localized mitochondria along with actin cables by following Cox4-DsRed and a GFP-tagged copy of the actin filament-binding protein, Abp140 (Yang and Pon, 2002
) (). Prior to cell-cell fusion, actin cables and mitochondria orient toward the zone of contact. Upon fusion, actin orientation becomes less obvious, the midzone often appears depleted of filaments, and – as detailed above – mitochondria extend only to the midpoint of the zygote. When buds become visible, actin has reorganized to generate cables that extend from the bud neck and extend either a) in roughly symmetric fashion toward each parental domain (when budding is medial), or b) along the long axis of the zygote (when budding is non-medial). In each case, mitochondria appear to align with cables.
In mitotic cells, septins and the formin, Bnr1p, localize to the bud neck and are required for nucleating linear actin filaments in the mother cell. A second formin, Bni1p, localizes to the bud tip and plays a similar role for organization of actin in the bud (Buttery et al., 2007
; Pruyne et al., 2004
). We therefore investigated the interdependence of septin localization, Bnr1p and actin polarization in zygotes.
In crosses between wildtype cells, we observe that GFP-tagged Bnr1p becomes visible only when zygotic buds emerge, i.e. approximately when actin becomes repolarized and parental mitochondria fuse. At this time it colocalizes with septins at the bud neck ().
To learn whether septins at the bud neck are required for colocalization of Bnr1p and for actin polarization throughout the zygote, we allowed bud formation to begin and then inactivated one septin at the bud neck. In these experiments, we formed cdc11-6 zygotes at 23°C, using a pair of temperature-sensitive strains that express mCherry-tagged Cdc3p and GFP-tagged Bnr1p. When the zygotes were then shifted to 36–37°C, both tagged Cdc3p and tagged Bnr1p disappeared from the bud neck (, panel 1). Moreover, in equivalent protocols in which one of the haploid parents expressed tagged Abp140, actin cables became dramatically concentrated in the elongated buds and remarkably absent from the rest of the zygote (, panels 2–4). No such changes occurred in parallel experiments with wildtype zygotes. Since bnr1Δ strains do not form conventional zygotes with good efficiency, it has not been possible to inquire whether Bnr1p itself is needed (unpublished observations).
There thus is a close connection between arrival of septins at the bud neck and the polarization of actin in the body of the zygote. We propose that this repolarization of actin facilitates fusion of parental mitochondria in zygotes.