Influenza viruses were propagated either in eggs or in Madin-Darby canine kidney (MDCK) cells. The A(H1N1)pdm09 viruses used included A/California/07/2009-PR8 (X-179A; vaccine virus) and egg- and cell-propagated 6B.1 and 6B.2 viruses, A/Michigan/45/2015 (MI/15e and MI/15c) and A/Iowa/53/2015 (IW/15e and IW/15c), respectively. Ten pre-2009 seasonal A(H1N1) (sH1N1) viruses representing major antigenic clusters circulated during 1977–2009 were also used: A/USSR/90/1977 (USSR/77), A/England/333/1980 (ENG/80), A/Chile/1/1983 (CHI/83), A/Taiwan/1/1986 (TW/86), A/Texas/36/1991(TX/91), A/Bayern/7/1995(BAY/95), A/Beijing/262/1995 (BJ/95), A/New Caledonia/20/1999 (NC/99), A/Solomon Islands/3/2006 (SI/06), and A/Brisbane/59/2007 (BR/07). In addition, 3 reverse genetically (RG) engineered egg-propagated viruses were generated, containing the HA and neuraminidase genes from CA/09 and 6 internal genes from A/Puerto Rico/8/1934, including A/California/7/2009-PR8 (RG-163K), A/California/7/2009-PR8-K163Q (RG-K163Q), and A/California/7/2009-PR8-D127T (RG-D127T) with a putative glycosylation motif at position 125. Details of RG virus generation are included in the Supplementary Material.

Prevaccination and 21–28-day postvaccination serum samples from US adults (n = 336) who received inactivated influenza vaccines (IIVs) containing CA/09-like H1 component were collected from 6 seasons after the 2009 A(H1N1) pandemic. These anonymized serum samples were provided by a contract organization, and their use was exempt from the Centers for Disease Control and Prevention (CDC) human subjects review. Postinfection ferret antisera were generated by intranasal inoculation of a single dose of wild-type viruses [16].

Hemagglutination inhibition (HI) assays were conducted with 4 HA units per 25 μL of virus, using 0.5% turkey erythrocytes, as described elsewhere [17]. Serum samples from low responders were further tested using microneutralization assays [17] to measure neutralizing antibody responses. In addition, a small subset of postvaccination serum samples also underwent antibody adsorption with purified whole viruses to confirm the specificity of antibodies [18, 19]. Details of the assays are described in the Supplementary Material.

Wilcoxon matched paired t tests were used to analyze geometric mean titers (GMTs). Fisher exact test was used to compare the numbers of low responders to A(H1N1)pdm09 viruses. Differences were considered significant at P < .05. GraphPad Prism V5 software (GraphPad Software) was used for statistical analyses.

Paired serum samples collected from vaccinated adults (aged 18–49 years; n = 336) in 6 seasons since the 2009 pandemic were first tested using HI assays against the X-179A virus. Samples from 300 individuals (aged 18–49 years; birth year [BY], 1961–1998; median age, 33 years) with postvaccination HI titer ≥40 to vaccine like viruses were further analyzed (Table 1 and Supplementary Table 1). The majority of vaccinees mounted robust antibody responses to X-179A (GMT, 272); 87%–98% had postvaccination HI titers ≥40 to other A(H1N1) pdm09 viruses (Table 1). However, compared with HI titers to egg-propagated X-179A across the 6 seasons (n = 300), 9% and 12% of vaccinees had ≥4 fold reductions in HI titers to egg-propagated circulating 6B.1 and 6B.2 viruses respectively; 22% and 25% of these individual also had ≥4 fold reductions in HI titers to cell-propagated 6B.1 and 6B.2 viruses, respectively (Table 1). Furthermore, there was a significant reduction in HI titers to RG-163Q (GMT, 136) compared with titers to RG-163K (GMT, 202) (P < .05) across 6 seasons and in every individual season (P < .05).

The HI GMTs to cell-propagated 6B.1 (GMT, 156) and 6B.2 (GMT, 115) viruses bearing 223Q were significantly lower (P < .05) than those to the corresponding egg-propagated 6B.1 (GMT, 226) and 6B.2 (GMT, 147) viruses bearing the egg-adapted Q223R change (Table 1). The only difference between the egg- and corresponding cell-propagated 6B.1 and 6B.2 viruses is the Q223R change (Supplementary Table 2). These results suggest that although IIV induced robust HI antibodies to the vaccine, 163K and 223R epitopes in the HA of the vaccine virus contributed to reduced antibody responses to circulating viruses (163Q, 223Q) in some adults.

Antibody responses to A(H1N1)pdm09 viruses can be shaped by prior exposure to sH1N1 viruses [4, 20–22]. To systemically identify the likely priming sH1N1 virus for each individual, paired serum samples were tested against 10 sH1N1 viruses representative of antigenic clusters circulated during 1977–2009 (Supplementary Figure 1) to which individuals could have been exposed since birth. Antigenic characterization using ferret antisera illustrated the antigenic drift of these pre-2009 viruses over time. Ferret antisera generated through A(H1N1)pdm09 primary infection do not inhibit any of the pre-2009 sH1N1 viruses (Table 2). In contrast, vaccination with CA/09-like virus in adults not only induced antibodies to the vaccine and circulating A(H1N1)pdm09 viruses; it also elicited recalled antibody responses to multiple pre-2009 sH1N1 viruses (Figure 1 and Supplementary Figure 2). Individuals with postvaccination HI titers ≥40 to ≥2 of the 10 sH1N1 viruses (n = 281) after vaccination were included in further analysis to define immune priming. Using each person’s antibody profile to pre-2009 historic sH1N1 viruses and BY, along with the epidemic periods of the sH1N1 viruses (Supplementary Figure 1B), and also taking into consideration the possible period of delay from birth to the first influenza priming exposure [8], we were able to construct 3 distinct age-related priming patterns to 1977–2009 sH1N1 viruses among study individuals.

A total of 112 individuals from the oldest cohort (BY, 1961–1983; median BY [MBY], 1972) elicited strong HI antibodies (titer ≥40) to USSR/77, ENG/80, or CHI/83 viruses after CA/09 vaccination. GMTs to these viruses were higher than to 1986–2009 sH1N1 viruses (Figure 1A). There was also a significant rise in GMTs to all the prior-2009 sH1N1 viruses with IIV vaccination, with the higher rate of seroconversion (52 of 112; 46%) being to USSR/77, ENG/80, and CHI/83 viruses; a lower proportion (≤27%) of individuals seroconverted to the 1986–2009 sH1N1 viruses. Collectively, based on these results with human serum samples (Figure 1A), antigenic characteristics from ferret antisera (Table 2) and epidemic periods of these viruses (1977–1986; Supplementary Figure 1), USSR/77, ENG/80, and CHI/83 viruses were designated as the USSR/77-like virus group. These 112 individuals were more than likely first infected with USSR/77-like viruses and thus designated as the USSR/77-primed cohort (Figure 2B).

In contrast, 154 individuals (BY, 1965–1996; MBY, 1985) showed uniformly similar HI titers to TW/86, TX/91, and BAY/95 viruses (GMT, 190–210) after vaccination. Postvaccination titers to these 3 viruses were significantly higher (≥4-fold) than titers to the other 7 sH1N1 viruses (P < .01) (Figure 1B and Supplementary Figure 1). The majority of this cohort showed undetectable HI titers to the 3 USSR/77-like viruses. Based on the responses with human serum samples (Figure 1B), ferret antisera (Table 2), and epidemic periods of these viruses (1986–2001; Supplementary Figure 1), TW/86, TX/91, and BAY/95 viruses were designated as TW/86-like virus group. These 154 individuals were probably first exposed to TW/86-like viruses and thus designated as the TW/86-primed cohort (Figure 2B).

Finally, after IIV vaccination, 15 individuals (BY, 1972–1998; MBY, 1994; 13 were born in 1990–1998) mounted the highest responses (HI titers, ≥40) to NC/99 virus (GMT, 88) or similar titers (≤2 fold differences) to SI/06, BJ/95, and BR/07 viruses which all contain a deletion at position 130 on the HA (Supplemental Figure 1), significantly higher (≥4 fold) than titers to TW/86-group viruses (P < .01). None showed detectable HI titers to USSR/77-like viruses. This cohort was probably first exposed to NC/99-like viruses after birth and thus designated as the NC/99-primed cohort (Figure 2B). Interestingly, significantly reduced GMTs to 6B.1 and 6B.2 viruses compared with those against X-179A, and significantly reduced GMTs to RG-K163Q and RG-D127T viruses compared with those against RG-K163, were noted only among the USSR-primed cohort (Figure 1).

Comparing HI titers to X-179A and titers to egg- and cell-propagated circulating 6B.1 and 6B.2 viruses after vaccination, we identified 3 groups of low responders: (1) individuals who mounted 163K-specific responses (defined as a ≥4-fold reduction in HI titers to RG-K163Q vs RG-163K virus); (2) those who had 223R-specific antibodies targeting the egg-adapted Q223R change (defined as ≥4-fold reduction in titers to cell-propagated 6B.1 and 6B.2 viruses compared with the corresponding egg-propagated viruses); and (3) those with both 163K+223R-specific antibodies (demonstrating both 1 and 2). Proportions of the low responders among the 3 priming cohorts (total N = 281) were analyzed for all 6 seasons (Figure 2A). A total of 54 low responders were identified, of whom 25 showed 223R-specific 25 showed 163K-specific and 4 showed 163K+223R-specific HI antibodies. Figure 2B illustrates the age and priming pattern distribution of the low responders among all 281 individuals analyzed. 223R-specific HI antibodies were detected in 7%, 10%, and 13% of the USSR/77-, TW/86- and NC/99-primed cohorts, respectively. In contrast, 163K-specific HI antibodies were detected only in the USSR/77-primed cohort (22% targeting 163K-specific epitope only and 4% recognizing epitopes involving both 163K+223R). Among the 54 low responders, 27 had HI titers to MI/15c virus (representing circulating viruses) below the protective threshold of 40, accounting for approximately 10% of the total cohorts (Figure 2C).

Significantly reduced HI titers to viruses with HA 163Q (n = 29) or 223Q (n = 25) epitopes after vaccination were also confirmed by reactivity of neutralizing antibodies (Figure 3). It is noteworthy that ferret antisera to A(H1N1)pdm09 viruses failed to detect antigenic changes caused by either K163Q or Q223R substitution (Table 2).

We then tested 163K-specific postvaccination human serum pools from 3 seasons with additional A(H1N1)pdm09 viruses circulated since 2009 from genetic groups 5, 6A, 6C, and 7 (163K) and additional 6B.1 and 6B.2 viruses (163Q). Human serum sample pools containing 163K-specific antibodies all had reduced antibody responses to 6B.1 and 6B.2 viruses (≥8 fold reduction compared with responses to egg- and cell-propagated CA/09), but not to the other A(H1N1)pdm09 genetic groups. These antigenic differences were not recognized by ferret antiserum to CA/09 (Table 3).

All 29 individuals showing 163K specificity were from the USSR/77-primed cohort (Figure 2). Ninety percent (26 of 29) seroconverted to X-179A (163K) and 81% (21/26) seroconverted to the 3 USSR/77-like viruses (163K) (Supplementary Figure 2B), indicating responses to shared dominant epitopes on these viruses. USSR/77-primed cohorts were probably imprinted with 163K-specific epitopes from first infection with a USSR/77-like virus (Supplementary Figure 2). In contrast, reduced titers and lower seroconversions were observed to both 1986–2009 sH1N1 (Supplementary Figure 2B) and RG-D127T viruses (Figure 3A and 3B). This is consistent with a previous study, which proposed that although 163K was also present in 1986–2009 sH1N1 viruses, glycosylation at position 125 could conformationally shield 163 and render 163K epitope less dominant when primed by these viruses [4]. It is noteworthy that the antibody response was very poor (HI titers, ≤20) to MI/15c (163Q) in 62% of USSR/77-primed individuals (18 of 29) (Supplementary Figure 2A), indicating that antibodies targeting 163K dominate in these persons. These data provide further immune basis for the age-specific low VE to A(H1N1) viruses observed in middle-aged adults [7, 8].

Finally, we explored the antibody repertoire in USSR/77-primed low responders using antibody adsorption techniques. Corresponding to the lower HI titers detected against 163Q viruses (Figure 4A), adsorption results for 3 representative USSR/77-primed individuals clearly illustrated that 2-way cross-reactivity was specifically observed between the USSR/77-group viruses (163K) and X-179A/RG-163K virus but not with MI/15e (163Q) and RG-K163Q viruses (Figure 4B), supporting the antigenic basis of shared 163K epitope. In addition, for 2 individuals (Figure 4A; I-2 and I-3) with undetectable or low titers to 223Q virus (MI/15c), serum adsorption with 223Q virus did not significantly reduce titers to X-179A and MI/15e (223R; Figure 4B; I-2 and I-3), confirming 223R specificity. In 1 individual (Figure 4; I-3), antibody to MI/15c was removed by adsorption with multiple viruses, suggesting the presence of cross-reactive antibodies that are not targeting 163 and 223 epitopes.